version 3.3.91

* NEWS: Record release date.

.gitignore

@@ -1,3 +1,4 @@
+/ABOUT-NLS~
 *.eps
 *.log
 *.o

.prev-version

@@ -1 +1 @@
-3.3
+3.3.90

.travis.yml

@@ -8,7 +8,8 @@ language: cpp
 env:
   global:
     # ICC serial number.
-    secure: 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
+    # See https://github.com/nemequ/icc-travis.
+    secure: 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
 
 matrix:
   include:
@@ -289,7 +290,7 @@ matrix:
       env:
         # No versioned name installed, but beware that Travis installs
         # a more modern clang earlier in the default PATH.
-        - MATRIX_EVAL='CC=/usr/bin/clang && CXX=/usr/bin/clang++'
+        - MATRIX_EVAL='CC=/usr/bin/clang && CXX=/usr/bin/clang++ && GETTEXT_VERSION=0.18.3'
 
     - name: "Clang 3.3"
       os: linux
@@ -301,7 +302,7 @@ matrix:
             - clang-3.3
       env:
         # See comment for 3.4.
-        - MATRIX_EVAL='CC=/usr/bin/clang && CXX=/usr/bin/clang++'
+        - MATRIX_EVAL='CC=/usr/bin/clang && CXX=/usr/bin/clang++ && GETTEXT_VERSION=0.18.3'
 
 
 before_install:
@@ -339,7 +340,10 @@ script:
   # with `Picked up _JAVA_OPTIONS: -Xmx2048m -Xms512m`, which makes
   # the test suite fail.
   - unset _JAVA_OPTIONS
+  - if [[ $GETTEXT_VERSION ]]; then sed -i -re "s/(AM_GNU_GETTEXT_VERSION).*/\1([$GETTEXT_VERSION])/" configure.ac; fi
   - ./bootstrap
+  # For some reason, gnulib-po/Makefile.in.in is about Gettext 0.19, so it will break here.  Override it.
+  - if [[ $GETTEXT_VERSION ]]; then autopoint --force; fi
   - if [[ -f ~/.bashrc ]]; then source ~/.bashrc; fi
   - ./configure --enable-gcc-warnings CC="$CC" CXX="$CXX" $CONFIGUREFLAGS || { cat config.log && false; }
   - make -j2 $MAKE_ARGS

Makefile.am

@@ -18,7 +18,7 @@
 AUTOMAKE_OPTIONS = subdir-objects
 ACLOCAL_AMFLAGS = -I m4
 
-SUBDIRS = po runtime-po .
+SUBDIRS = po runtime-po gnulib-po .
 
 # Files installed for use by Automake.
 aclocaldir = @aclocaldir@
@@ -29,12 +29,12 @@ EXTRA_DIST = .prev-version .version cfg.mk PACKAGING \
 
 dist_doc_DATA = AUTHORS COPYING NEWS README THANKS TODO
 
-# Let generated files (e.g., from exampkles/) depend on this variable,
+# Let generated files (e.g., from examples/) depend on this variable,
 # to make sure they are regenerated when there are changes in Bison.
 #
 # Don't depend on $(BISON) otherwise we would rebuild these files
 # in srcdir, including during distcheck, which is forbidden.
-dependencies = $(BISON_IN) $(dist_skeletons_DATA)
+dependencies = $(BISON_IN) $(src_bison_SOURCES) $(dist_skeletons_DATA)
 
 ## Running the bison from this tarball.  To generate our own parser,
 ## but also to run the tests.  Of course, you ought to keep a sane
@@ -42,7 +42,8 @@ dependencies = $(BISON_IN) $(dist_skeletons_DATA)
 BISON = $(top_builddir)/tests/bison
 BISON_IN = $(top_srcdir)/tests/bison.in
 YACC = $(BISON) -o y.tab.c
-AM_YFLAGS = -d -v -Werror -Wall -Wno-yacc --report=all
+AM_YFLAGS_WITH_LINES = --defines -Werror -Wall --report=all
+AM_YFLAGS = $(AM_YFLAGS_WITH_LINES) --no-lines
 
 # Initialization before completion by local.mk's.
 AM_CFLAGS = $(WARN_CFLAGS)
@@ -87,9 +88,22 @@ clean-local:
 BUILT_SOURCES += $(top_srcdir)/.version
 $(top_srcdir)/.version: configure
 	echo $(VERSION) > $@-t && mv $@-t $@
-dist-hook: gen-ChangeLog
+dist-hook: gen-ChangeLog gen-synclines
 	echo $(VERSION) > $(distdir)/.tarball-version
 
+# When generating a release, include the #lines in Bison's parser.  Do it
+# only if needed.  In particular, distcheck runs dist with read-only
+# sources, so don't try to update src/parse-gram.c: we don't have the
+# permissions.
+.PHONY: gen-synclines
+gen-synclines:
+	cd $(distdir) &&						\
+	if ! grep '#line' src/parse-gram.c >/dev/null 2>&1; then	\
+	  $(abs_top_builddir)/tests/bison				\
+	    $(AM_YFLAGS_WITH_LINES) src/parse-gram.y			\
+	    -o src/parse-gram.c;					\
+	fi
+
 .PHONY: update-b4-copyright update-package-copyright-year
 update-b4-copyright:
 	$(AM_V_GEN)find $(srcdir)/data -type f          \

NEWS

@@ -1,5 +1,150 @@
 GNU Bison NEWS
 
+* Noteworthy changes in release 3.3.91 (2019-05-12) [beta]
+
+** Bug fixes
+
+  Portability issues were fixed.
+
+* Noteworthy changes in release 3.3.90 (2019-04-28) [beta]
+
+** New features
+
+*** Colored diagnostics
+
+  As an experimental feature, Bison now generates colored diagnostics,
+  controlled by the new options --color and --style.
+
+  Install the libtextstyle library before configuring Bison to use them.
+  It is available from
+
+    https://alpha.gnu.org/gnu/gettext/
+
+  for instance
+
+    https://alpha.gnu.org/gnu/gettext/libtextstyle-0.8.tar.gz
+
+  The option --color supports the following arguments:
+    - always, yes: Enable colors.
+    - never, no: Disable colors.
+    - auto, tty (default): Enable colors if the output device is a tty.
+
+  To customize the styles, create a CSS file similar to
+
+    /* black-and-white.css */
+    .warning   { }
+    .error     { font-weight: 800; text-decoration: underline; }
+    .note      { }
+
+  then invoke bison with --style=black-and-white.css.
+
+*** Disabling output
+
+  When given -fsyntax-only, the diagnostics are reported, but no output is
+  generated.
+
+  The name of this option is somewhat misleading as bison does more than
+  just checking the syntax: every stage is run (including checking for
+  conflicts for instance), except the generation of the output files.
+
+*** Include the generated header (yacc.c)
+
+  Before, when --defines is used, bison generated a header, and pasted an
+  exact copy of it into the generated parser implementation file.  If the
+  header name is not "y.tab.h", it is now #included instead of being
+  duplicated.
+
+  To use an '#include' even if the header name is "y.tab.h" (which is what
+  happens with --yacc, or when using the Autotools' ylwrap), define
+  api.header.include to the exact argument to pass to #include.  For
+  instance:
+
+    %define api.header.include {"parse.h"}
+
+  or
+
+    %define api.header.include {<parser/parse.h>}
+
+*** api.location.type is now supported in C (yacc.c, glr.c)
+
+  The %define variable api.location.type defines the name of the type to use
+  for locations.  When defined, Bison no longer defines YYLTYPE.
+
+  This can be used in programs with several parsers to factor their
+  definition of locations: let one of them generate them, and the others
+  just use them.
+
+** Changes
+
+*** Graphviz output
+
+  In conformance with the recommendations of the Graphviz team, if %require
+  "3.4" (or better) is specified, the option --graph generates a *.gv file
+  by default, instead of *.dot.
+
+*** Diagnostics overhaul
+
+  Column numbers were wrong with multibyte characters, which would also
+  result in skewed diagnostics with carets.  Beside, because we were
+  indenting the quoted source with a single space, lines with tab characters
+  were incorrectly underlined.
+
+  To address these issues, and to be clearer, Bison now issues diagnostics
+  as GCC9 does.  For instance it used to display (there's a tab before the
+  opening brace):
+
+    foo.y:3.37-38: error: $2 of ‘expr’ has no declared type
+     expr: expr '+' "number"        { $$ = $1 + $2; }
+                                         ^~
+  It now reports
+
+    foo.y:3.37-38: error: $2 of ‘expr’ has no declared type
+        3 | expr: expr '+' "number" { $$ = $1 + $2; }
+          |                                     ^~
+
+  Other constructs now also have better locations, resulting in more precise
+  diagnostics.
+
+*** Fix-it hints for %empty
+
+  Running Bison with -Wempty-rules and --update will remove incorrect %empty
+  annotations, and add the missing ones.
+
+*** Generated reports
+
+  The format of the reports (parse.output) was improved for readability.
+
+*** Better support for --no-line.
+
+  When --no-line is used, the generated files are now cleaner: no lines are
+  generated instead of empty lines.  Together with using api.header.include,
+  that should help people saving the generated files into version control
+  systems get smaller diffs.
+
+** Documentation
+
+  A new example in C shows an simple infix calculator with a hand-written
+  scanner (examples/c/calc).
+
+  A new example in C shows a reentrant parser (capable of recursive calls)
+  built with Flex and Bison (examples/c/reccalc).
+
+  There is a new section about the history of Yaccs and Bison.
+
+** Bug fixes
+
+  A few obscure bugs were fixed, including the second oldest (known) bug in
+  Bison: it was there when Bison was entered in the RCS version control
+  system, in December 1987.  See the NEWS of Bison 3.3 for the previous
+  oldest bug.
+
+* Noteworthy changes in release 3.3.2 (2019-02-03) [stable]
+
+** Bug fixes
+
+  Bison 3.3 failed to generate parsers for grammars with unused nonterminal
+  symbols.
+
 * Noteworthy changes in release 3.3.1 (2019-01-27) [stable]
 
 ** Changes
@@ -216,17 +361,18 @@ GNU Bison NEWS
     symbol_type (int token, const int&);
     symbol_type (int token);
 
-  which should be used in a Flex-scanner as follows.
-
-    %%
-    [a-z]+   return yy::parser::symbol_type (ID, yytext);
-    [0-9]+   return yy::parser::symbol_type (INT, text_to_int (yytext);
-    ":"      return yy::parser::symbol_type (’:’);
-    <<EOF>>  return yy::parser::symbol_type (0);
-
   Correct matching between token types and value types is checked via
-  'assert'.  For instance, 'symbol_type (ID, 42)' would abort (while
-  'make_ID (42)' would not even compile).
+  'assert'; for instance, 'symbol_type (ID, 42)' would abort.  Named
+  constructors are preferable, as they offer better type safety (for
+  instance 'make_ID (42)' would not even compile), but symbol_type
+  constructors may help when token types are discovered at run-time, e.g.,
+
+     [a-z]+   {
+                if (auto i = lookup_keyword (yytext))
+                  return yy::parser::symbol_type (i);
+                else
+                  return yy::parser::make_ID (yytext);
+              }
 
 *** C++: Variadic emplace
 
@@ -3479,7 +3625,7 @@ along with this program.  If not, see <http://www.gnu.org/licenses/>.
  LocalWords:  Heimbigner AST src ast Makefile srcdir MinGW xxlex XXSTYPE
  LocalWords:  XXLTYPE strictfp IDEs ffixit fdiagnostics parseable fixits
  LocalWords:  Wdeprecated yytext Variadic variadic yyrhs yyphrs RCS README
- LocalWords:  noexcept constexpr ispell american
+ LocalWords:  noexcept constexpr ispell american deprecations
 
 Local Variables:
 ispell-dictionary: "american"

README

@@ -1,7 +1,18 @@
 This package contains the GNU Bison parser generator.
 
 * Installation
-** Build
+** Build from git
+Here are basic installation instructions for a repository checkout:
+
+    $ git submodules update --init
+    $ ./bootstrap
+
+then proceed with the usual 'configure && make' steps.
+
+README-hacking contains more information about building and modifying the
+software.
+
+** Build for tarball
 See the file INSTALL for generic compilation and installation instructions.
 
 Bison requires GNU m4 1.4.6 or later.  See:

README-hacking

@@ -133,9 +133,15 @@ Enjoy!
 
 ** Updating
 
-The use of submodules make things somewhat different because git does not
-support recursive operations: submodules must be taken care of explicitly by
-the user.
+If you have git at version 1.8.2 or later, the command
+
+        $ git submodule update --recursive --remote
+
+will be useful for updating to the latest version of all submodules.
+
+Under earlier versions, use of submodules make things somewhat different
+because git does not yet support recursive operations: submodules must be
+taken care of explicitly.
 
 *** Updating Bison
 
@@ -189,6 +195,11 @@ update.
 ** make check
 Use liberally.
 
+** Updating the expectations
+Sometimes some changes have a large impact on the test suite (e.g., when we
+added the "[-Wother]" part to all the warnings).  Part of the update can be
+done with a crude tool: tests/update-test.  Read it for more information.
+
 ** TESTSUITEFLAGS
 
 To run just the testsuite (not the tests related to the examples), run `make
@@ -282,6 +293,24 @@ release:
 - Test with a very recent version of GCC for both C and C++.  Testing with
   older versions that are still in use is nice too.
 
+** gnulib
+To run tests on gnulib components (e.g., on bitset):
+
+    cd gnulib
+    ./gnulib-tool --test bitset-tests
+
+possibly within a specified environment:
+
+    CC='gcc-mp-8 -fsanitize=undefined' ./gnulib-tool --test bitset-tests
+
+To be able to run the tests several times, and to use symlinks instead of
+copies so that one can update the origin gnulib directory and immediately
+re-run the tests, run:
+
+    ./gnulib-tool --symlink --create-test --dir=/tmp/gnutest bitset-tests
+    cd /tmp/gnutest
+    ./configure CC='gcc-mp-8 -fsanitize=undefined'
+    make check
 
 * Release Procedure
 This section needs to be updated to take into account features from gnulib.

REFERENCES

@@ -1,33 +0,0 @@
-From phr Tue Jul  8 10:36:19 1986
-Date: Tue, 8 Jul 86 00:52:24 EDT
-From: phr (Paul Rubin)
-To: riferguson%[email protected], tower
-Subject: Re:  Bison documentation?
-
-The main difference between Bison and Yacc that I know of is that
-Bison supports the @N construction, which gives you access to
-the starting and ending line number and character number associated
-with any of the symbols in the current rule.
-
-Also, Bison supports the command '%expect N' which says not to mention
-the conflicts if there are N shift/reduce conflicts and no reduce/reduce
-conflicts.
-
-The differences in the algorithms stem mainly from the horrible
-kludges that Johnson had to perpetrate to make Yacc fit in a PDP-11.
-
-Also, Bison uses a faster but less space-efficient encoding for the
-parse tables (see Corbett's PhD thesis from Berkeley, "Static
-Semantics in Compiler Error Recovery", June 1985, Report No. UCB/CSD
-85/251), and more modern technique for generating the lookahead sets.
-(See Frank DeRemer and Thomas Pennello, "Efficient Computation of
-LALR(1) Look-Ahead Sets", ACM Transactions on Programming Languages
-and Systems (TOPLAS) 4, 4 (October 1982), 615-649.  Their
-technique is the standard one now.)
-
-        paul rubin
-        free software foundation
-
-
-[DeRemer-Pennello reference corrected by Paul Eggert <[email protected]>,
- 2004-06-21.]

THANKS

@@ -18,6 +18,7 @@ Antonio Silva Correia     [email protected]
 Arnold Robbins            [email protected]
 Art Haas                  [email protected]
 Askar Safin               [email protected]
+Balázs Scheidler          [email protected]
 Baron Schwartz            [email protected]
 Ben Pfaff                 [email protected]
 Benoit Perrot             [email protected]
@@ -55,6 +56,7 @@ Didier Godefroy           [email protected]
 Efi Fogel                 [email protected]
 Enrico Scholz             [email protected]
 Eric Blake                [email protected]
+Eric S. Raymond           [email protected]
 Étienne Renault           [email protected]
 Evgeny Stambulchik        [email protected]
 Fabrice Bauzac            [email protected]
@@ -64,6 +66,7 @@ Frank Heckenbach          [email protected]
 Frans Englich             [email protected]
 Gabriel Rassoul           [email protected]
 Gary L Peskin             [email protected]
+Gavin Smith               [email protected]
 Georg Sauthoff            [email protected]
 George Neuner             [email protected]
 Gilles Espinasse          [email protected]
@@ -126,6 +129,7 @@ Oliver Mangold            [email protected]
 Paolo Bonzini             [email protected]
 Paolo Simone Gasparello   [email protected]
 Pascal Bart               [email protected]
+Patrice Dumas             [email protected]
 Paul Eggert               [email protected]
 Paul Hilfinger            [email protected]
 Per Allansson             [email protected]
@@ -168,12 +172,14 @@ Tim Josling               [email protected]
 Tim Landscheidt           [email protected]
 Tim Van Holder            [email protected]
 Tobias Frost              [email protected]
+Todd Freed                [email protected]
 Tom Lane                  [email protected]
 Tom Tromey                [email protected]
 Tommy Nordgren            [email protected]
 Troy A. Johnson           [email protected]
 Tys Lefering              [email protected]
 Valentin Tolmer           [email protected]
+wcventure                 [email protected]
 Victor Khomenko           [email protected]
 Victor Zverovich          [email protected]
 Vin Shelton               [email protected]

TODO

@@ -1,7 +1,37 @@
 * Bison 3.4
-** injection rules
+** bad diagnostics
+%token <val> NUM
+%type  <val> expr term fact
+
+%%
+
+res:  expr { printf ("%d\n", $1); };
+expr: expr '+' term { $$ = $1 + $3; } | term;
+term: NUM | { $$ = 0; };
+
+The second warning about fact is... useless.
+
+$ bison /tmp/bar.y
+/tmp/bar.y:2.24-27: warning: symbol fact is used, but is not defined as a token and has no rules [-Wother]
+ %type  <val> expr term fact
+                        ^~~~
+/tmp/bar.y: warning: 1 nonterminal useless in grammar [-Wother]
+/tmp/bar.y:2.24-27: warning: nonterminal useless in grammar: fact [-Wother]
+ %type  <val> expr term fact
+                        ^~~~
+
+* Bison 3.5
+** doc
+I feel its ugly to use the GNU style to declare functions in the doc.  It
+generates tons of white space in the page, and may contribute to bad page
+breaks.
+
+Also, we seem to teach YYPRINT very early on, although it should be
+considered deprecated: %printer is superior.
+
 ** glr.cc
 move glr.c into the yy namespace
+
 ** improve syntax errors (UTF-8, internationalization)
 Bison depends on the current locale.  For instance:
 
@@ -45,15 +75,50 @@ syntax error, unexpected $end, expecting ↦ or 🎅🐃 or '\n'
 While at it, we should stop using "$end" by default, in favor of "end of
 file", or "end of input", whatever.
 
-** use gettext-h in gnulib instead of gettext
-** use gnulib-po
-For some reason, according to syntax-check, we have to keep getopt.c in
-POTIFILES.in, but not bitset/stats.c, although both come from gnulib.  But
-bitset/stats.c is a symlink, not getopt.c.  This is fishy and should be
-fixed.
+* Bison 3.6
+** Unit rules
+Maybe we could expand unit rules (or "injections", see
+https://homepages.cwi.nl/~daybuild/daily-books/syntax/2-sdf/sdf.html), i.e.,
+transform
+
+        exp: arith | bool;
+        arith: exp '+' exp;
+        bool: exp '&' exp;
+
+into
+
+        exp: exp '+' exp | exp '&' exp;
+
+when there are no actions.  This can significantly speed up some grammars.
+I can't find the papers.  In particular the book 'LR parsing: Theory and
+Practice' is impossible to find, but according to 'Parsing Techniques: a
+Practical Guide', it includes information about this issue.  Does anybody
+have it?
+
+** Injection rules
+See above.
+
+** clean up
+*** lalr.c
+Introduce a goto struct, and use it in place of from_state/to_state.
+Rename states1 as path, length as pathlen.
+Introduce inline functions for things such as nullable[*rp - ntokens]
+where we need to map from symbol number to nterm number.
+
+There are probably a significant part of the relations management that
+should be migrated on top of a bitsetv.
+
+*** closure
+It should probably take a "state*" instead of two arguments.
+
+*** traces
+The "automaton" and "set" categories are not so useful.  We should probably
+introduce lr(0) and lalr, just the way we have ielr categories.  The
+"closure" function is too verbose, it should probably have its own category.
+
+"set" can still be used for summariring the important sets.  That would make
+tests easy to maintain.
 
-Meanwhile, bitset/stats.c is removed from the set of translations, which is
-not too much of a problem as users are not expected to see this.
 
 
 * Completion
@@ -77,10 +142,6 @@ $ ./tests/testsuite -l | grep errors | sed q
 ** consistency
 token vs terminal
 
-** yacc.c
-Now that ylwrap is fixed, we should include foo.tab.h from foo.tab.c rather
-than duplicating it.
-
 ** C++
 Move to int everywhere instead of unsigned?  stack_size, etc.  The parser
 itself uses int (for yylen for instance), yet stack is based on size_t.
@@ -171,9 +232,6 @@ class in the printers/destructors, which is not good for an operator<<
 since it is no longer bound to a particular parser, it's just a
 (standalone symbol).
 
-** Rename LR0.cc
-as lr0.cc, why upper case?
-
 * Various
 ** Rewrite glr.cc in C++
 As a matter of fact, it would be very interesting to see how much we can
@@ -263,7 +321,7 @@ grammars she is working on.  We should probably also include some
 information about the variables (I'm not sure for instance we even
 specify what LR variant was used).
 
-**  GLR
+** GLR
 How would Paul like to display the conflicted actions?  In particular,
 what when two reductions are possible on a given lookahead token, but one is
 part of $default.  Should we make the two reductions explicit, or just
@@ -355,33 +413,6 @@ XML output for GNU Bison
 https://lists.gnu.org/archive/html/bug-bison/2016-06/msg00000.html
 http://www.cs.cornell.edu/andru/papers/cupex/
 
-* Unit rules
-Maybe we could expand unit rules (or "injections", see
-https://homepages.cwi.nl/~daybuild/daily-books/syntax/2-sdf/sdf.html), i.e.,
-transform
-
-        exp: arith | bool;
-        arith: exp '+' exp;
-        bool: exp '&' exp;
-
-into
-
-        exp: exp '+' exp | exp '&' exp;
-
-when there are no actions.  This can significantly speed up some grammars.
-I can't find the papers.  In particular the book 'LR parsing: Theory and
-Practice' is impossible to find, but according to 'Parsing Techniques: a
-Practical Guide', it includes information about this issue.  Does anybody
-have it?
-
-
-
-* Documentation
-
-** History/Bibliography
-Some history of Bison and some bibliography would be most welcome.
-Are there any Texinfo standards for bibliography?
-
 * Coding system independence
 Paul notes:
 

bootstrap.conf

@@ -17,17 +17,18 @@
 
 # gnulib modules used by this package.
 gnulib_modules='
-  argmatch array-list assert
+  argmatch array-list assert assure
   bitsetv
   calloc-posix close closeout config-h c-strcase
   configmake
   dirname
   error extensions fdl fopen-safer
   getopt-gnu
-  gettext git-version-gen gitlog-to-changelog
+  gettext-h git-version-gen gitlog-to-changelog
   gpl-3.0 hash inttypes isnan javacomp-script
   javaexec-script
   ldexpl
+  libtextstyle-optional
   malloc-gnu
   mbswidth
   non-recursive-gnulib-prefix-hack
@@ -67,24 +68,7 @@ XGETTEXT_OPTIONS_RUNTIME=$XGETTEXT_OPTIONS'\\\
  --keyword=YY_ \\\
 '
 
-# Gettext supplies these files, but we don't need them since
-# we don't have an intl subdirectory.
-excluded_files='
-    m4/glibc2.m4
-    m4/intdiv0.m4
-    m4/intl.m4
-    m4/intldir.m4
-    m4/intmax.m4
-    m4/lcmessage.m4
-    m4/lock.m4
-    m4/longdouble.m4
-    m4/signed.m4
-    m4/uintmax_t.m4
-    m4/ulonglong.m4
-    m4/visibility.m4
-'
-
-gnulib_tool_option_extras='--symlink --conditional-dependencies --makefile-name=gnulib.mk'
+gnulib_tool_option_extras='--symlink --conditional-dependencies --makefile-name=gnulib.mk --po-base=gnulib-po --po-domain=bison'
 
 bootstrap_post_import_hook()
 {

build-aux/.gitignore

@@ -27,3 +27,4 @@
 /install-reloc
 /reloc-ldflags
 /relocatable.sh.in
+/libtool-reloc

build-aux/update-test

@@ -0,0 +1,93 @@
+#! /usr/bin/env python
+
+# usage:
+# update-test _build/8d/tests/testsuite.dir/*.testsuite.log
+
+import argparse
+import os
+import re
+
+
+def getargs():
+    p = argparse.ArgumentParser(description='Update test cases.')
+    opt = p.add_argument
+    opt('tests', metavar='test', nargs='+', type=str, default=None,
+        help='test files to update')
+    opt('-v', '--verbose', action='store_true',
+        help='Be verbose')
+    return p.parse_args()
+
+args = getargs()
+subst = dict()
+
+
+def log(*args_):
+    if args.verbose:
+        print(*args_)
+
+
+def contents(file):
+    '''The contents of a file.'''
+    log(file)
+    f = open(file)
+    return f.read()
+
+
+def diff_to_re(match):
+    '''Convert a portion of patch into a regex substitution to perform.
+    No longer used, we now use the expected/effective parts.
+    '''
+    frm = []
+    to = []
+    is_diff = False
+    for l in match.group(1).splitlines():
+        print(l)
+        # t in [-+ ]
+        t = l[0]
+        l = l[1:]
+        if t in ['-', ' ']:
+            is_diff = True
+            frm.append(l)
+        if t in ['+', ' ']:
+            is_diff = True
+            to.append(l)
+    if is_diff:
+        frm = "\n".join(frm)
+        to = "\n".join(to)
+        subst[frm] = to
+
+
+def update(at_file, logfile):
+    test = contents(at_file)
+    if os.path.isfile(logfile):
+        log("LOG: ", logfile)
+        l = contents(logfile)
+        log("LOG: ", l)
+        global subst
+        subst = {}
+        re.sub(r'(?:^@@.*\n)((?:^[-+ ].*\n)+)',
+               diff_to_re, l, flags = re.MULTILINE)
+        print(subst)
+        if subst:
+            # Turn "subst{frm} -> to" into a large RE.
+            frm = '|'.join([re.escape(x) for x in subst])
+            log("FROM:", frm)
+            test = re.sub("(" + frm + ")",
+                          lambda m: subst[m.group(1)],
+                          test, flags=re.MULTILINE)
+    open(at_file, 'w').write(test)
+
+
+def process(logfile):
+    log = contents(logfile)
+    # Look for the file to update.
+    m = re.search(r'^\d+\. (\w+\.at):\d+: ', log, re.MULTILINE)
+    if not m:
+        return
+    at_file = 'tests/' + m.group(1)
+    print(at_file)
+    update(at_file, logfile)
+
+for t in args.tests:
+    log("FILE:", t)
+    process(t)

cfg.mk

@@ -60,18 +60,6 @@ update-copyright-env = \
 ## More syntax-checks.  ##
 ## -------------------- ##
 
-# At least for Mac OS X's grep, the order between . and [ in "[^.[]"
-# matters:
-# $ LC_ALL=fr_FR grep -nE '[^[.]' /dev/null
-# $ LC_ALL=C grep -nE '[^[.]' /dev/null
-# grep: invalid collating element or class
-# $ LC_ALL=fr_FR grep -nE '[^.[]' /dev/null
-# $ LC_ALL=C grep -nE '[^.[]' /dev/null
-sc_at_parser_check:
-	@prohibit='AT_PARSER_CHECK\(\[+[^.[]|AT_CHECK\(\[+\./'		\
-	halt='use AT_PARSER_CHECK for and only for generated parsers'	\
-	  $(_sc_search_regexp)
-
 # Indent only with spaces.
 # Taken from Coreutils.
 sc_prohibit_tab_based_indentation:
@@ -156,11 +144,16 @@ sc_space_before_open_paren:
 ## syntax-checks exceptions.  ##
 ## -------------------------- ##
 
+# po-check: we use gnulib-po, so we don't need/want them in our POTFILE.
+generated_files =
+
 exclude = \
   $(foreach a,$(1),$(eval $(subst $$,$$$$,exclude_file_name_regexp--sc_$(a))))
+
 $(call exclude,                                                                 \
   bindtextdomain=^lib/main.c$$                                                  \
   cast_of_argument_to_free=^src/muscle-tab.c$$                                  \
+  po_check=^po/POTFILES.in$$                                                    \
   preprocessor_indentation=^data/|^lib/|^src/parse-gram.[ch]$$                  \
   program_name=^lib/main.c$$                                                    \
   prohibit_always-defined_macros=^data/skeletons/yacc.c$$                       \
@@ -169,8 +162,8 @@ $(call exclude,
   prohibit_doubled_word=^tests/named-refs.at$$                                  \
   prohibit_magic_number_exit=^doc/bison.texi$$                                  \
   prohibit_magic_number_exit+=?|^tests/(conflicts|regression).at$$              \
-  prohibit_strcmp=^doc/bison\.texi|tests/local\.at$$                            \
-  prohibit_tab_based_indentation=install-icc.sh|\.(am|mk)$$|^\.git|Makefile$$   \
+  prohibit_strcmp=^doc/bison\.texi|examples|tests/local\.at$$                   \
+  prohibit_tab_based_indentation=install-icc.sh|\.(am|mk)$$|^\.git|tests/input.at|Makefile$$   \
   require_config_h=^(lib/yyerror|data/skeletons/(glr|yacc))\.c$$                \
   require_config_h_first=^(lib/yyerror|data/skeletons/(glr|yacc))\.c$$          \
   space_before_open_paren=^data/skeletons/                                      \

configure.ac

@@ -192,6 +192,7 @@ BISON_CXX_COMPILER_POSIXLY_CORRECT
 
 # D.
 AC_CHECK_PROGS([DC], [dmd])
+AC_CHECK_PROGS([DCFLAGS], [])
 AM_CONDITIONAL([ENABLE_D], [test x"$DC" != x])
 
 # Java.
@@ -248,8 +249,9 @@ AM_WITH_DMALLOC
 # We use gnulib, which is only guaranteed to work properly with the
 # latest Gettext.
 AM_GNU_GETTEXT([external], [need-ngettext])
-AM_GNU_GETTEXT_VERSION([0.18])
+AM_GNU_GETTEXT_VERSION([0.19])
 BISON_I18N
+AC_CONFIG_FILES([gnulib-po/Makefile.in])
 
 # Internationalized parsers.
 AC_CONFIG_FILES([runtime-po/Makefile.in])

data/README.md

@@ -75,48 +75,76 @@ skeletons.
 
 ## Symbols
 
+### `b4_symbol(NUM, FIELD)`
 In order to unify the handling of the various aspects of symbols (tag, type
 name, whether terminal, etc.), bison.exe defines one macro per (token,
 field), where field can `has_id`, `id`, etc.: see
-src/output.c:prepare_symbols_definitions().
+`prepare_symbols_definitions()` in `src/output.c`.
 
-The various FIELDS are:
+The macro `b4_symbol(NUM, FIELD)` gives access to the following FIELDS:
+
+- `has_id`: 0 or 1.
 
-- has_id: 0 or 1.
   Whether the symbol has an id.
-- id: string
-  If has_id, the id.  Guaranteed to be usable as a C identifier.
-  Prefixed by api.token.prefix if defined.
-- tag: string.
+
+- `id`: string
+  If has_id, the id (prefixed by api.token.prefix if defined), otherwise
+  defined as empty.  Guaranteed to be usable as a C identifier.
+
+- `tag`: string.
   A representation of the symbol.  Can be 'foo', 'foo.id', '"foo"' etc.
-- user_number: integer
+
+- `user_number`: integer
   The external number as used by yylex.  Can be ASCII code when a character,
   some number chosen by bison, or some user number in the case of
   %token FOO <NUM>.  Corresponds to yychar in yacc.c.
-- is_token: 0 or 1
+
+- `is_token`: 0 or 1
   Whether this is a terminal symbol.
-- number: integer
+
+- `number`: integer
   The internal number (computed from the external number by yytranslate).
   Corresponds to yytoken in yacc.c.  This is the same number that serves as
   key in b4_symbol(NUM, FIELD).
-- has_type: 0, 1
+
+  In bison, symbols are first assigned increasing numbers in order of
+  appearance (but tokens first, then nterms).  After grammar reduction,
+  unused nterms are then renumbered to appear last (i.e., first tokens, then
+  used nterms and finally unused nterms).  This final number NUM is the one
+  contained in this field, and it is the one used as key in `b4_symbol(NUM,
+  FIELD)`.
+
+  The code of the rule actions, however, is emitted before we know what
+  symbols are unused, so they use the original numbers.  To avoid confusion,
+  they actually use "orig NUM" instead of just "NUM".  bison also emits
+  definitions for `b4_symbol(orig NUM, number)` that map from original
+  numbers to the new ones.  `b4_symbol` actually resolves `orig NUM` in the
+  other case, i.e., `b4_symbol(orig 42, tag)` would return the tag of the
+  symbols whose original number was 42.
+
+- `has_type`: 0, 1
   Whether has a semantic value.
-- type_tag: string
+
+- `type_tag`: string
   When api.value.type=union, the generated name for the union member.
   yytype_INT etc. for symbols that has_id, otherwise yytype_1 etc.
-- type
+
+- `type`
   If it has a semantic value, its type tag, or, if variant are used,
   its type.
   In the case of api.value.type=union, type is the real type (e.g. int).
-- has_printer: 0, 1
-- printer: string
-- printer_file: string
-- printer_line: integer
+
+- `has_printer`: 0, 1
+- `printer`: string
+- `printer_file`: string
+- `printer_line`: integer
+- `printer_loc`: location
   If the symbol has a printer, everything about it.
-- has_destructor, destructor, destructor_file, destructor_line
+
+- `has_destructor`, `destructor`, `destructor_file`, `destructor_line`, `destructor_loc`
   Likewise.
 
-### b4_symbol_value(VAL, [SYMBOL-NUM], [TYPE-TAG])
+### `b4_symbol_value(VAL, [SYMBOL-NUM], [TYPE-TAG])`
 Expansion of $$, $1, $<TYPE-TAG>3, etc.
 
 The semantic value from a given VAL.
@@ -127,14 +155,14 @@ The semantic value from a given VAL.
 The result can be used safely, it is put in parens to avoid nasty precedence
 issues.
 
-### b4_lhs_value(SYMBOL-NUM, [TYPE])
+### `b4_lhs_value(SYMBOL-NUM, [TYPE])`
 Expansion of `$$` or `$<TYPE>$`, for symbol `SYMBOL-NUM`.
 
-### b4_rhs_data(RULE-LENGTH, POS)
+### `b4_rhs_data(RULE-LENGTH, POS)`
 The data corresponding to the symbol `#POS`, where the current rule has
 `RULE-LENGTH` symbols on RHS.
 
-### b4_rhs_value(RULE-LENGTH, POS, SYMBOL-NUM, [TYPE])
+### `b4_rhs_value(RULE-LENGTH, POS, SYMBOL-NUM, [TYPE])`
 Expansion of `$<TYPE>POS`, where the current rule has `RULE-LENGTH` symbols
 on RHS.
 

data/diagnostics.css

@@ -0,0 +1,10 @@
+/* CSS style for Bison's diagnostics.
+
+   This is an experimental feature.  The class names may change in the
+   future.  */
+
+.warning   { color: purple; }
+.error     { color: red; }
+.note      { color: cyan; }
+/* Semantic values in Bison's parser traces.  */
+.value     { color: green; }

data/local.mk

@@ -15,7 +15,8 @@
 ## along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
 dist_pkgdata_DATA =                             \
-  data/README
+  data/README.md                                \
+  data/diagnostics.css
 
 skeletonsdir = $(pkgdatadir)/skeletons
 dist_skeletons_DATA =                           \

data/skeletons/bison.m4

@@ -389,17 +389,28 @@ m4_define([b4_glr_cc_if],
 #
 # The following macros provide access to symbol related values.
 
-# _b4_symbol(NUM, FIELD)
-# ----------------------
+# __b4_symbol(NUM, FIELD)
+# -----------------------
 # Recover a FIELD about symbol #NUM.  Thanks to m4_indir, fails if
 # undefined.
-m4_define([_b4_symbol],
+m4_define([__b4_symbol],
 [m4_indir([b4_symbol($1, $2)])])
 
 
+# _b4_symbol(NUM, FIELD)
+# ----------------------
+# Recover a FIELD about symbol #NUM (or "orig NUM").  Fails if
+# undefined.
+m4_define([_b4_symbol],
+[m4_ifdef([b4_symbol($1, number)],
+          [__b4_symbol(m4_indir([b4_symbol($1, number)]), $2)],
+          [__b4_symbol([$1], [$2])])])
+
+
+
 # b4_symbol(NUM, FIELD)
 # ---------------------
-# Recover a FIELD about symbol #NUM.  Thanks to m4_indir, fails if
+# Recover a FIELD about symbol #NUM (or "orig NUM").  Fails if
 # undefined.  If FIELD = id, prepend the token prefix.
 m4_define([b4_symbol],
 [m4_case([$2],
@@ -427,13 +438,6 @@ m4_define([b4_symbol_tag_comment],
 ])
 
 
-# b4_symbol_action_location(SYMBOL-NUM, KIND)
-# -------------------------------------------
-# Report the location of the KIND action as FILE:LINE.
-m4_define([b4_symbol_action_location],
-[b4_symbol([$1], [$2_file]):b4_syncline([b4_symbol([$1], [$2_line])])])
-
-
 # b4_symbol_action(SYMBOL-NUM, KIND)
 # ----------------------------------
 # Run the action KIND (destructor or printer) for SYMBOL-NUM.
@@ -444,9 +448,9 @@ m4_define([b4_symbol_action],
                    [],
                    [(*yylocationp)])dnl
     _b4_symbol_case([$1])[]dnl
-b4_syncline([b4_symbol([$1], [$2_line])], [b4_symbol([$1], [$2_file])])
+b4_syncline([b4_symbol([$1], [$2_line])], [b4_symbol([$1], [$2_file])])dnl
       b4_symbol([$1], [$2])
-b4_syncline([@oline@], [@ofile@])
+b4_syncline([@oline@], [@ofile@])dnl
         break;
 
 b4_dollar_popdef[]dnl
@@ -575,12 +579,15 @@ m4_define([b4_basename],
 [m4_bpatsubst([$1], [^.*/\([^/]+\)/*$], [\1])])
 
 
-# b4_syncline(LINE, FILE)
-# -----------------------
+# b4_syncline(LINE, FILE)dnl
+# --------------------------
+# Should always be following by a dnl.
+#
+# Emit "#line LINE FILE /* __LINE__ __FILE__ */".
 m4_define([b4_syncline],
 [b4_flag_if([synclines],
-[b4_sync_start([$1], [$2]) b4_sync_end([__line__],
-                                       [b4_basename(m4_quote(__file__))])[]dnl
+[b4_sync_start([$1], [$2])[]dnl
+b4_sync_end([__line__], [b4_basename(m4_quote(__file__))])
 ])])
 
 # b4_sync_start(LINE, FILE)
@@ -592,7 +599,17 @@ m4_define([b4_sync_start], [b4_comment([$2:$1])])
 # -----------------------
 # Syncline for the current place, which ends.  Typically a comment
 # left for the reader.
-m4_define([b4_sync_end],   [b4_comment([$2:$1])])
+m4_define([b4_sync_end],   [ b4_comment([$2:$1])]
+)
+# This generates dependencies on the Bison skeletons hence lots of
+# useless 'git diff'.  This location is useless for the regular
+# user (who does not care about the skeletons) and is actually not
+# useful for Bison developpers too (I, Akim, never used this to locate
+# the code in skeletons that generated output).  So disable it
+# completely.  If someone thinks this was actually useful, a %define
+# variable should be provided to control the level of verbosity of
+# '#line', in replacement of --no-lines.
+m4_define([b4_sync_end])
 
 
 # b4_user_code(USER-CODE)
@@ -742,8 +759,10 @@ m4_define([b4_percent_define_get_kind],
           [m4_indir([b4_percent_define_kind(]$1[)])],
           [b4_fatal([[$0: undefined %%define variable '%s']], [$1])])])
 
-# b4_percent_define_get_syncline(VARIABLE)
-# ----------------------------------------
+# b4_percent_define_get_syncline(VARIABLE)dnl
+# -------------------------------------------
+# Should always be following by a dnl.
+#
 # Mimic muscle_percent_define_get_syncline in ../src/muscle-tab.h exactly.
 # That is, if the %define variable VARIABLE is undefined, complain fatally
 # since that's a Bison or skeleton error.  Otherwise, return its definition
@@ -959,8 +978,7 @@ m4_ifval([$1], [m4_define([b4_percent_code_bison_qualifiers(]$1[)])])dnl
 m4_ifdef(b4_macro_name,
 [b4_comment([m4_if([$#], [0], [[Unqualified %code]],
                    [["%code ]$1["]])[ blocks.]])
-b4_user_code([m4_indir(b4_macro_name)])
-])dnl
+b4_user_code([m4_indir(b4_macro_name)])])dnl
 m4_popdef([b4_macro_name])])
 
 # b4_percent_code_ifdef(QUALIFIER, IF-TRUE, [IF-FALSE])
@@ -1060,10 +1078,10 @@ m4_define_default([b4_location_initial_line],   [1])
 ## Sanity checks.  ##
 ## --------------- ##
 
-# api.location.prefix={...} (Java and C++).
+# api.location.type={...} (C, C++ and Java).
 b4_percent_define_check_kind([api.location.type], [code], [deprecated])
 
-# api.position.prefix={...} (Java).
+# api.position.type={...} (Java).
 b4_percent_define_check_kind([api.position.type], [code], [deprecated])
 
 # api.prefix >< %name-prefix.

data/skeletons/c++.m4

@@ -147,13 +147,13 @@ m4_if(m4_bregexp(b4_namespace_ref, [::[	 ]*$]), [-1], [],
                 [[namespace reference has a trailing "::"]])])
 
 m4_define([b4_namespace_open],
-[b4_user_code([b4_percent_define_get_syncline([[api.namespace]])
+[b4_user_code([b4_percent_define_get_syncline([[api.namespace]])dnl
 [namespace ]m4_bpatsubst(m4_dquote(m4_bpatsubst(m4_dquote(b4_namespace_ref),
                                                 [^\(.\)[	 ]*::], [\1])),
                          [::], [ { namespace ])[ {]])])
 
 m4_define([b4_namespace_close],
-[b4_user_code([b4_percent_define_get_syncline([[api.namespace]])
+[b4_user_code([b4_percent_define_get_syncline([[api.namespace]])dnl
 m4_bpatsubst(m4_dquote(m4_bpatsubst(m4_dquote(b4_namespace_ref[ ]),
                                     [^\(.\)[	 ]*\(::\)?\([^][:]\|:[^:]\)*],
                                     [\1])),
@@ -194,7 +194,7 @@ m4_define([b4_value_type_declare],
 [union\|union-directive],
 [[    union semantic_type
     {
-    ]b4_user_union_members[
+]b4_user_union_members[
     };]])])dnl
 ])
 

data/skeletons/c.m4

@@ -229,18 +229,22 @@ m4_define([b4_attribute_define],
 #endif
 
 ]m4_bmatch([$1], [\bnoreturn\b], [[/* The _Noreturn keyword of C11.  */
-#if ! defined _Noreturn
-# if defined __cplusplus && 201103L <= __cplusplus
+]dnl This is an exact copy of lib/_Noreturn.h.
+[#ifndef _Noreturn
+# if (defined __cplusplus \
+      && ((201103 <= __cplusplus && !(__GNUC__ == 4 && __GNUC_MINOR__ == 7)) \
+          || (defined _MSC_VER && 1900 <= _MSC_VER)))
 #  define _Noreturn [[noreturn]]
-# elif !(defined __STDC_VERSION__ && 201112 <= __STDC_VERSION__)
-#  if (3 <= __GNUC__ || (__GNUC__ == 2 && 8 <= __GNUC_MINOR__) \
-       || 0x5110 <= __SUNPRO_C)
-#   define _Noreturn __attribute__ ((__noreturn__))
-#  elif defined _MSC_VER && 1200 <= _MSC_VER
-#   define _Noreturn __declspec (noreturn)
-#  else
-#   define _Noreturn
-#  endif
+# elif ((!defined __cplusplus || defined __clang__) \
+        && (201112 <= (defined __STDC_VERSION__ ? __STDC_VERSION__ : 0)  \
+            || 4 < __GNUC__ + (7 <= __GNUC_MINOR__)))
+   /* _Noreturn works as-is.  */
+# elif 2 < __GNUC__ + (8 <= __GNUC_MINOR__) || 0x5110 <= __SUNPRO_C
+#  define _Noreturn __attribute__ ((__noreturn__))
+# elif 1200 <= (defined _MSC_VER ? _MSC_VER : 0)
+#  define _Noreturn __declspec (noreturn)
+# else
+#  define _Noreturn
 # endif
 #endif
 
@@ -470,7 +474,7 @@ m4_define([b4_sync_start], [[#]line $1 $2])
 m4_define([b4_case],
 [  case $1:
 $2
-b4_syncline([@oline@], [@ofile@])
+b4_syncline([@oline@], [@ofile@])dnl
     break;])
 
 
@@ -480,7 +484,7 @@ m4_define([b4_predicate_case],
 [  case $1:
     if (! (
 $2)) YYERROR;
-b4_syncline([@oline@], [@ofile@])
+b4_syncline([@oline@], [@ofile@])dnl
     break;])
 
 
@@ -540,8 +544,10 @@ m4_if(b4_skeleton, ["yacc.c"],
     YYPRINT (yyo, yytoknum[yytype], *yyvaluep);
 # endif
 ]])dnl
-  b4_symbol_actions([printer])[
-}
+b4_percent_code_get([[pre-printer]])dnl
+  b4_symbol_actions([printer])
+b4_percent_code_get([[post-printer]])dnl
+[}
 
 
 /*---------------------------.
@@ -709,13 +715,13 @@ typedef ]b4_percent_define_get([[api.value.type]])[ ]b4_api_PREFIX[STYPE;
 [m4_bmatch(b4_percent_define_get([[api.value.type]]),
 [union\|union-directive],
 [[#if ! defined ]b4_api_PREFIX[STYPE && ! defined ]b4_api_PREFIX[STYPE_IS_DECLARED
-]b4_percent_define_get_syncline([[api.value.union.name]])[
-union ]b4_percent_define_get([[api.value.union.name]])[
+]b4_percent_define_get_syncline([[api.value.union.name]])dnl
+[union ]b4_percent_define_get([[api.value.union.name]])[
 {
 ]b4_user_union_members[
 };
-]b4_percent_define_get_syncline([[api.value.union.name]])[
-typedef union ]b4_percent_define_get([[api.value.union.name]])[ ]b4_api_PREFIX[STYPE;
+]b4_percent_define_get_syncline([[api.value.union.name]])dnl
+[typedef union ]b4_percent_define_get([[api.value.union.name]])[ ]b4_api_PREFIX[STYPE;
 # define ]b4_api_PREFIX[STYPE_IS_TRIVIAL 1
 # define ]b4_api_PREFIX[STYPE_IS_DECLARED 1
 #endif
@@ -726,7 +732,10 @@ typedef union ]b4_percent_define_get([[api.value.union.name]])[ ]b4_api_PREFIX[S
 # -----------------------
 m4_define([b4_location_type_define],
 [[/* Location type.  */
-#if ! defined ]b4_api_PREFIX[LTYPE && ! defined ]b4_api_PREFIX[LTYPE_IS_DECLARED
+]b4_percent_define_ifdef([[api.location.type]],
+[[typedef ]b4_percent_define_get([[api.location.type]])[ ]b4_api_PREFIX[LTYPE;
+]],
+[[#if ! defined ]b4_api_PREFIX[LTYPE && ! defined ]b4_api_PREFIX[LTYPE_IS_DECLARED
 typedef struct ]b4_api_PREFIX[LTYPE ]b4_api_PREFIX[LTYPE;
 struct ]b4_api_PREFIX[LTYPE
 {
@@ -738,7 +747,7 @@ struct ]b4_api_PREFIX[LTYPE
 # define ]b4_api_PREFIX[LTYPE_IS_DECLARED 1
 # define ]b4_api_PREFIX[LTYPE_IS_TRIVIAL 1
 #endif
-]])
+]])])
 
 
 # b4_declare_yylstype

data/skeletons/d.m4

@@ -18,11 +18,24 @@
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
 
-# b4_comment(TEXT)
-# ----------------
-m4_define([b4_comment], [/* m4_bpatsubst([$1], [
-], [
-   ])  */])
+# _b4_comment(TEXT, OPEN, CONTINUE, END)
+# --------------------------------------
+# Put TEXT in comment.  Avoid trailing spaces: don't indent empty lines.
+# Avoid adding indentation to the first line, as the indentation comes
+# from OPEN.  That's why we don't patsubst([$1], [^\(.\)], [   \1]).
+#
+# Prefix all the output lines with PREFIX.
+m4_define([_b4_comment],
+[$2[]m4_bpatsubst(m4_expand([[$1]]), [
+\(.\)], [
+$3\1])$4])
+
+
+# b4_comment(TEXT, [PREFIX])
+# --------------------------
+# Put TEXT in comment.  Prefix all the output lines with PREFIX.
+m4_define([b4_comment],
+[_b4_comment([$1], [$2/* ], [$2   ], [  */])])
 
 
 # b4_list2(LIST1, LIST2)
@@ -177,7 +190,7 @@ b4_percent_define_default([[stype]], [[YYSemanticType]])])
 # %name-prefix
 m4_define_default([b4_prefix], [[YY]])
 
-b4_percent_define_default([[api.parser.class]], [b4_prefix[]YYParser])])
+b4_percent_define_default([[api.parser.class]], [b4_prefix[]Parser])])
 m4_define([b4_parser_class], [b4_percent_define_get([[api.parser.class]])])
 
 #b4_percent_define_default([[location_type]], [Location])])

data/skeletons/glr.c

@@ -196,14 +196,14 @@ m4_if(b4_skeleton, ["glr.c"],
 # glr.cc produces its own header.
 b4_glr_cc_if([],
 [b4_defines_if(
-[b4_output_begin([b4_spec_defines_file])
+[b4_output_begin([b4_spec_header_file])
 b4_copyright([Skeleton interface for Bison GLR parsers in C],
              [2002-2015, 2018-2019])[
-]b4_cpp_guard_open([b4_spec_defines_file])[
+]b4_cpp_guard_open([b4_spec_header_file])[
 ]b4_shared_declarations[
-]b4_cpp_guard_close([b4_spec_defines_file])[
-]b4_output_end
-])])
+]b4_cpp_guard_close([b4_spec_header_file])[
+]b4_output_end[
+]])])
 
 
 # ------------------------- #
@@ -238,7 +238,7 @@ b4_copyright([Skeleton implementation for Bison GLR parsers in C],
 
 ]b4_null_define[
 
-]b4_defines_if([[#include "@basename(]b4_spec_defines_file[@)"]],
+]b4_defines_if([[#include "@basename(]b4_spec_header_file[@)"]],
                [b4_shared_declarations])[
 
 /* Enabling verbose error messages.  */

data/skeletons/glr.cc

@@ -100,8 +100,8 @@ m4_defn([b4_initial_action])]))])[
 
 # Hijack the post prologue to declare yyerror.
 ]m4_append([b4_post_prologue],
-[b4_syncline([@oline@], [@ofile@])[
-]b4_function_declare([yyerror],
+[b4_syncline([@oline@], [@ofile@])dnl
+b4_function_declare([yyerror],
     [static void],b4_locations_if([
     [[const ]b4_namespace_ref::b4_parser_class[::location_type *yylocationp],
                         [yylocationp]],])
@@ -128,8 +128,8 @@ m4_if(b4_prefix, [yy], [],
 # Hijack the epilogue to define implementations (yyerror, parser member
 # functions etc.).
 m4_append([b4_epilogue],
-[b4_syncline([@oline@], [@ofile@])[
-
+[b4_syncline([@oline@], [@ofile@])dnl
+[
 /*------------------.
 | Report an error.  |
 `------------------*/
@@ -234,7 +234,7 @@ m4_pushdef([b4_parse_param], m4_defn([b4_parse_param_orig]))dnl
 
 #endif
 ]m4_popdef([b4_parse_param])dnl
-b4_namespace_close
+b4_namespace_close[]dnl
 ])
 
 
@@ -351,15 +351,15 @@ b4_percent_define_flag_if([[global_tokens_and_yystype]],
 ])
 
 b4_defines_if(
-[b4_output_begin([b4_spec_defines_file])
+[b4_output_begin([b4_spec_header_file])
 b4_copyright([Skeleton interface for Bison GLR parsers in C++],
              [2002-2015, 2018-2019])[
 // C++ GLR parser skeleton written by Akim Demaille.
 
 ]b4_disclaimer[
-]b4_cpp_guard_open([b4_spec_defines_file])[
+]b4_cpp_guard_open([b4_spec_header_file])[
 ]b4_shared_declarations[
-]b4_cpp_guard_close([b4_spec_defines_file])[
+]b4_cpp_guard_close([b4_spec_header_file])[
 ]b4_output_end])
 
 # Let glr.c (and b4_shared_declarations) believe that the user

data/skeletons/lalr1.cc

@@ -111,9 +111,9 @@ b4_dollar_pushdef([yysym.value],
                   [],
                   [yysym.location])dnl
       _b4_symbol_case([$1])[]dnl
-b4_syncline([b4_symbol([$1], [$2_line])], [b4_symbol([$1], [$2_file])])
+b4_syncline([b4_symbol([$1], [$2_line])], [b4_symbol([$1], [$2_file])])dnl
         b4_symbol([$1], [$2])
-b4_syncline([@oline@], [@ofile@])
+b4_syncline([@oline@], [@ofile@])dnl
         break;
 
 m4_popdef([b4_symbol_value])[]dnl
@@ -399,22 +399,22 @@ m4_define([b4_shared_declarations],
 ## -------------- ##
 
 b4_defines_if(
-[b4_output_begin([b4_spec_defines_file])
+[b4_output_begin([b4_spec_header_file])
 b4_copyright([Skeleton interface for Bison LALR(1) parsers in C++])
 [
 /**
- ** \file ]b4_spec_defines_file[
+ ** \file ]b4_spec_header_file[
  ** Define the ]b4_namespace_ref[::parser class.
  */
 
 // C++ LALR(1) parser skeleton written by Akim Demaille.
 
 ]b4_disclaimer[
-]b4_cpp_guard_open([b4_spec_defines_file])[
+]b4_cpp_guard_open([b4_spec_header_file])[
 ]b4_shared_declarations(hh)[
-]b4_cpp_guard_close([b4_spec_defines_file])
-b4_output_end
-])
+]b4_cpp_guard_close([b4_spec_header_file])[
+]b4_output_end[
+]])
 
 
 b4_output_begin([b4_parser_file_name])[
@@ -428,7 +428,7 @@ m4_if(b4_prefix, [yy], [],
 
 ]b4_user_pre_prologue[
 
-]b4_defines_if([[#include "@basename(]b4_spec_defines_file[@)"]],
+]b4_defines_if([[#include "@basename(]b4_spec_header_file[@)"]],
                [b4_shared_declarations([cc])])[
 
 ]b4_user_post_prologue[

data/skeletons/lalr1.d

@@ -47,8 +47,8 @@ import std.format;
  * parser <tt>]b4_parser_class[</tt>.
  */
 public interface Lexer
-{
-  ]b4_locations_if([[/**
+{]b4_locations_if([[
+  /**
    * Method to retrieve the beginning position of the last scanned token.
    * @@return the position at which the last scanned token starts.  */
   @@property ]b4_position_type[ startPos ();
@@ -56,8 +56,9 @@ public interface Lexer
   /**
    * Method to retrieve the ending position of the last scanned token.
    * @@return the first position beyond the last scanned token.  */
-  @@property ]b4_position_type[ endPos ();]])[
+  @@property ]b4_position_type[ endPos ();
 
+]])[
   /**
    * Method to retrieve the semantic value of the last scanned token.
    * @@return the semantic value of the last scanned token.  */
@@ -68,13 +69,13 @@ public interface Lexer
    * to the next token and prepares to return the semantic value
    * ]b4_locations_if([and beginning/ending positions ])[of the token.
    * @@return the token identifier corresponding to the next token. */
-  YYTokenType yylex ();
+  int yylex ();
 
   /**
    * Entry point for error reporting.  Emits an error
    * ]b4_locations_if([referring to the given location ])[in a user-defined way.
-   *
-   * ]b4_locations_if([[@@param loc The location of the element to which the
+   *]b4_locations_if([[
+   * @@param loc The location of the element to which the
    *                error message is related]])[
    * @@param s The string for the error message.  */
    void yyerror (]b4_locations_if([b4_location_type[ loc, ]])[string s);
@@ -82,8 +83,8 @@ public interface Lexer
 
 private final struct YYStackElement{
   int state;
-  ]b4_yystype[ value;
-  ]b4_locations_if(b4_location_type[[] location;])[
+  ]b4_yystype[ value;]b4_locations_if(
+  b4_location_type[[] location;])[
 }
 
 private final struct YYStack {
@@ -231,9 +232,9 @@ b4_user_union_members
   ]b4_identification[
 
   /** True if verbose error messages are enabled.  */
-  public bool errorVerbose = ]b4_flag_value([error_verbose]);
+  public bool errorVerbose = ]b4_flag_value([error_verbose])[;
 
-  b4_locations_if([[
+]b4_locations_if([[
   private final ]b4_location_type[ yylloc_from_stack (ref YYStack rhs, int n)
   {
     static if (yy_location_is_class) {
@@ -249,34 +250,34 @@ b4_user_union_members
     }
   }]])[
 
-  ]b4_lexer_if([[private class YYLexer implements Lexer {
+]b4_lexer_if([[  private class YYLexer implements Lexer {
 ]b4_percent_code_get([[lexer]])[
   }
-
-  ]])[/** The object doing lexical analysis for us.  */
+]])[
+  /** The object doing lexical analysis for us.  */
   private Lexer yylexer;
-  ]
-  b4_parse_param_vars
 
-b4_lexer_if([[
+]b4_parse_param_vars[
+
+]b4_lexer_if([[
   /**
    * Instantiates the Bison-generated parser.
    */
   public this] (b4_parse_param_decl([b4_lex_param_decl])[) {
     this.yylexer = new YYLexer(]b4_lex_param_call[);
     this.yyDebugStream = stderr;
-    ]b4_parse_param_cons[
+]b4_parse_param_cons[
   }
-]])
+]])[
 
   /**
    * Instantiates the Bison-generated parser.
    * @@param yylexer The scanner that will supply tokens to the parser.
    */
-  b4_lexer_if([[protected]], [[public]]) [this (]b4_parse_param_decl([[Lexer yylexer]])[) {
+  ]b4_lexer_if([[protected]], [[public]]) [this (]b4_parse_param_decl([[Lexer yylexer]])[) {
     this.yylexer = yylexer;
     this.yyDebugStream = stderr;
-    ]b4_parse_param_cons[
+]b4_parse_param_cons[
   }
 
   private File yyDebugStream;
@@ -358,8 +359,8 @@ b4_lexer_if([[
 
   private int yyaction (int yyn, ref YYStack yystack, int yylen)
   {
-    ]b4_yystype[ yyval;
-    ]b4_locations_if([b4_location_type[ yyloc = yylloc_from_stack (yystack, yylen);]])[
+    ]b4_yystype[ yyval;]b4_locations_if([[
+    ]b4_location_type[ yyloc = yylloc_from_stack (yystack, yylen);]])[
 
     /* If YYLEN is nonzero, implement the default value of the action:
        `$$ = $1'.  Otherwise, use the top of the stack.
@@ -408,7 +409,8 @@ b4_lexer_if([[
     if (yystr[0] == '"')
       {
         string yyr;
-        strip_quotes: for (int i = 1; i < yystr.length; i++)
+      strip_quotes:
+        for (int i = 1; i < yystr.length; i++)
           switch (yystr[i])
             {
             case '\'':
@@ -475,15 +477,15 @@ b4_locations_if([, ref ]b4_location_type[ yylocationp])[)
     YYStack yystack;
 
     /* Error handling.  */
-    int yynerrs_ = 0;
-    ]b4_locations_if([/// The location where the error started.
+    int yynerrs_ = 0;]b4_locations_if([[
+    /// The location where the error started.
     ]b4_location_type[ yyerrloc = null;
 
     /// ]b4_location_type[ of the lookahead.
     ]b4_location_type[ yylloc;
 
     /// @@$.
-    ]b4_location_type[ yyloc;])[
+    ]b4_location_type[ yyloc;]])[
 
     /// Semantic value of the lookahead.
     ]b4_yystype[ yylval;
@@ -531,8 +533,7 @@ m4_popdef([b4_at_dollar])])dnl
         if (yychar == yyempty_)
         {
           yycdebug ("Reading a token: ");
-          yychar = yylex ();]
-          b4_locations_if([[
+          yychar = yylex ();]b4_locations_if([[
           static if (yy_location_is_class) {
             yylloc = new ]b4_location_type[(yylexer.startPos, yylexer.endPos);
           } else {
@@ -624,7 +625,7 @@ m4_popdef([b4_at_dollar])])dnl
           yyerror (]b4_locations_if([yylloc, ])[yysyntax_error (yystate, yytoken));
         }
 
-        ]b4_locations_if([yyerrloc = yylloc;])[
+]b4_locations_if([        yyerrloc = yylloc;])[
         if (yyerrstatus_ == 3)
         {
           /* If just tried and failed to reuse lookahead token after an
@@ -648,9 +649,8 @@ m4_popdef([b4_at_dollar])])dnl
       /*-------------------------------------------------.
       | errorlab -- error raised explicitly by YYERROR.  |
       `-------------------------------------------------*/
-      case YYERROR:
-
-        ]b4_locations_if([yyerrloc = yystack.locationAt (yylen - 1);])[
+      case YYERROR:]b4_locations_if([
+        yyerrloc = yystack.locationAt (yylen - 1);])[
         /* Do not reclaim the symbols of the rule which action triggered
            this YYERROR.  */
         yystack.pop (yylen);
@@ -663,7 +663,7 @@ m4_popdef([b4_at_dollar])])dnl
       | yyerrlab1 -- common code for both syntax error and YYERROR.  |
       `-------------------------------------------------------------*/
       case YYERRLAB1:
-        yyerrstatus_ = 3;	/* Each real token shifted decrements this.xb  */
+        yyerrstatus_ = 3;       /* Each real token shifted decrements this.  */
 
         for (;;)
         {
@@ -683,14 +683,14 @@ m4_popdef([b4_at_dollar])])dnl
           if (yystack.height == 1)
             return false;
 
-          ]b4_locations_if([yyerrloc = yystack.locationAt (0);])[
+]b4_locations_if([          yyerrloc = yystack.locationAt (0);])[
           yystack.pop ();
           yystate = yystack.stateAt (0);
           if (yydebug > 0)
             yystack.print (yyDebugStream);
         }
 
-        ]b4_locations_if([
+]b4_locations_if([
         /* Muck with the stack to setup for yylloc.  */
         yystack.push (0, yy_semantic_null, yylloc);
         yystack.push (0, yy_semantic_null, yyerrloc);
@@ -892,6 +892,5 @@ m4_popdef([b4_at_dollar])])dnl
 b4_percent_code_get[]dnl
 
 }
-
-b4_epilogue
+b4_epilogue[]dnl
 m4_divert_pop(0)dnl

data/skeletons/lalr1.java

@@ -17,8 +17,7 @@
 
 m4_include(b4_skeletonsdir/[java.m4])
 
-b4_defines_if([b4_fatal([%s: %%defines does not make sense in Java],
-              [b4_skeleton])])
+b4_defines_if([b4_complain([%defines does not make sense in Java])])
 
 # We do not depend on %debug in Java, but pacify warnings about
 # non-used flags.
@@ -26,9 +25,8 @@ b4_parse_trace_if([0], [0])
 
 m4_define([b4_symbol_no_destructor_assert],
 [b4_symbol_if([$1], [has_destructor],
-              [b4_fatal([%s: %s: %%destructor does not make sense in Java],
-                        [b4_skeleton],
-                        [b4_symbol_action_location([$1], [destructor])])])])
+              [b4_complain_at(m4_unquote(b4_symbol([$1], [destructor_loc])),
+                              [%destructor does not make sense in Java])])])
 b4_symbol_foreach([b4_symbol_no_destructor_assert])
 
 # Setup some macros for api.push-pull.
@@ -89,8 +87,8 @@ b4_output_begin([b4_parser_file_name])[
 ]])[
 ]b4_user_pre_prologue[
 ]b4_user_post_prologue[
-]b4_percent_code_get([[imports]])
-[/**
+]b4_percent_code_get([[imports]])[
+/**
  * A Bison parser, automatically generated from <tt>]m4_bpatsubst(b4_file_name, [^"\(.*\)"$], [\1])[</tt>.
  *
  * @@author LALR (1) parser skeleton written by Paolo Bonzini.

data/skeletons/yacc.c

@@ -284,20 +284,48 @@ m4_define([b4_declare_yyparse],
 # Declaration that might either go into the header (if --defines)
 # or open coded in the parser body.
 m4_define([b4_shared_declarations],
-[b4_cpp_guard_open([b4_spec_defines_file])[
+[b4_cpp_guard_open([b4_spec_header_file])[
 ]b4_declare_yydebug[
 ]b4_percent_code_get([[requires]])[
 ]b4_token_enums_defines[
 ]b4_declare_yylstype[
 ]b4_declare_yyparse[
 ]b4_percent_code_get([[provides]])[
-]b4_cpp_guard_close([b4_spec_defines_file])[]dnl
+]b4_cpp_guard_close([b4_spec_header_file])[]dnl
 ])
 
+
+# b4_header_include_if(IF-TRUE, IF-FALSE)
+# ---------------------------------------
+# Run IF-TRUE if we generate an output file and api.header.include
+# is defined.
+m4_define([b4_header_include_if],
+[m4_ifval(m4_quote(b4_spec_header_file),
+          [b4_percent_define_ifdef([[api.header.include]],
+                                   [$1],
+                                   [$2])],
+          [$2])])
+
+m4_if(b4_spec_header_file, [[y.tab.h]],
+      [b4_percent_define_default([[api.header.include]],
+                                 [["@basename(]b4_spec_header_file[@)"]])])
+
+
+
+
 ## -------------- ##
 ## Output files.  ##
 ## -------------- ##
 
+
+b4_defines_if([[
+]b4_output_begin([b4_spec_header_file])[
+]b4_copyright([Bison interface for Yacc-like parsers in C])[
+]b4_disclaimer[
+]b4_shared_declarations[
+]b4_output_end[
+]])# b4_defines_if
+
 b4_output_begin([b4_parser_file_name])[
 ]b4_copyright([Bison implementation for Yacc-like parsers in C])[
 /* C LALR(1) parser skeleton written by Richard Stallman, by
@@ -345,11 +373,11 @@ m4_if(b4_api_prefix, [yy], [],
 # define YYERROR_VERBOSE ]b4_error_verbose_if([1], [0])[
 #endif
 
-]m4_ifval(m4_quote(b4_spec_defines_file),
-[[/* In a future release of Bison, this section will be replaced
-   by #include "@basename(]b4_spec_defines_file[@)".  */
-]])dnl
-b4_shared_declarations[
+]b4_header_include_if([[#include ]b4_percent_define_get([[api.header.include]])],
+                      [m4_ifval(m4_quote(b4_spec_header_file),
+                                [/* Use api.header.include to #include this header
+   instead of duplicating it here.  */
+])b4_shared_declarations])[
 
 ]b4_user_post_prologue[
 ]b4_percent_code_get[]dnl
@@ -411,6 +439,15 @@ typedef short yytype_int16;
 
 ]b4_attribute_define[
 
+]b4_parse_assert_if([[#ifdef NDEBUG
+# define YY_ASSERT(E) ((void) (0 && (E)))
+#else
+# include <assert.h> /* INFRINGES ON USER NAME SPACE */
+# define YY_ASSERT(E) assert (E)
+#endif
+]],
+[[#define YY_ASSERT(E) ((void) (0 && (E)))]])[
+
 #if ]b4_lac_if([[1]], [[! defined yyoverflow || YYERROR_VERBOSE]])[
 
 /* The parser invokes alloca or malloc; define the necessary symbols.  */]dnl
@@ -1453,6 +1490,8 @@ yynewstate:
 | yynewstate -- set current state (the top of the stack) to yystate.  |
 `--------------------------------------------------------------------*/
 yysetstate:
+  YYDPRINTF ((stderr, "Entering state %d\n", yystate));
+  YY_ASSERT (0 <= yystate && yystate < YYNSTATES);
   *yyssp = (yytype_int16) yystate;
 
   if (yyss + yystacksize - 1 <= yyssp)
@@ -1520,8 +1559,6 @@ yysetstate:
     }
 #endif /* !defined yyoverflow && !defined YYSTACK_RELOCATE */
 
-  YYDPRINTF ((stderr, "Entering state %d\n", yystate));
-
   if (yystate == YYFINAL)
     YYACCEPT;
 
@@ -1613,8 +1650,8 @@ yyread_pushed_token:]])[
   yystate = yyn;
   YY_IGNORE_MAYBE_UNINITIALIZED_BEGIN
   *++yyvsp = yylval;
-  YY_IGNORE_MAYBE_UNINITIALIZED_END
-]b4_locations_if([  *++yylsp = yylloc;])[
+  YY_IGNORE_MAYBE_UNINITIALIZED_END]b4_locations_if([
+  *++yylsp = yylloc;])[
   goto yynewstate;
 
 
@@ -1662,7 +1699,7 @@ yyreduce:
   }]], [[
   switch (yyn)
     {
-      ]b4_user_actions[
+]b4_user_actions[
       default: break;
     }]])[
   /* User semantic actions sometimes alter yychar, and that requires
@@ -1917,11 +1954,3 @@ yypushreturn:]])[
 }
 ]b4_epilogue[]dnl
 b4_output_end
-
-b4_defines_if([[
-]b4_output_begin([b4_spec_defines_file])[
-]b4_copyright([Bison interface for Yacc-like parsers in C])[
-]b4_disclaimer[
-]b4_shared_declarations[
-]b4_output_end[
-]])# b4_defines_if

doc/local.mk

@@ -14,7 +14,12 @@
 ## You should have received a copy of the GNU General Public License
 ## along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
-AM_MAKEINFOFLAGS = --no-split
+AM_MAKEINFOFLAGS =                                              \
+  --no-split                                                    \
+  --set-customization-variable=SECTION_NAME_IN_TITLE=true       \
+  --set-customization-variable=AVOID_MENU_REDUNDANCY=true       \
+  --set-customization-variable=ICONS=true
+
 info_TEXINFOS = doc/bison.texi
 doc_bison_TEXINFOS =                            \
   $(CROSS_OPTIONS_TEXI)                         \
@@ -28,7 +33,7 @@ doc_bison = doc/bison
 $(doc_bison).dvi: $(FIGS_GV:.gv=.eps)
 $(doc_bison).info: $(FIGS_GV:.gv=.txt)
 $(doc_bison).pdf: $(FIGS_GV:.gv=.pdf)
-$(doc_bison).html: $(FIGS_GV:.gv=.png)
+$(doc_bison).html: $(FIGS_GV:.gv=.svg)
 
 TEXI2DVI = texi2dvi --build-dir=doc/bison.t2d -I doc
 CLEANDIRS += doc/bison.t2d
@@ -88,11 +93,13 @@ EXTRA_DIST += $(top_srcdir)/doc/bison.help
 if ! CROSS_COMPILING
 MAINTAINERCLEANFILES += $(top_srcdir)/doc/bison.help
 $(top_srcdir)/doc/bison.help: src/bison$(EXEEXT)
-	$(AM_V_GEN)LC_ALL=C src/bison$(EXEEXT) --version >doc/bison.help.tmp
-	$(AM_V_at) LC_ALL=C src/bison$(EXEEXT) --help | \
+	$(AM_V_GEN)LC_ALL=C tests/bison --version >doc/bison.help.tmp
+	$(AM_V_at) LC_ALL=C tests/bison --help | \
+## Avoid depending on the path to Bison.
+	  sed -e 's,^Usage: .*/bison \[OPTION\],Usage: bison [OPTION],g' \
 ## Avoid variations in the output depending on whether we are
 ## on a glibc system.
-	  sed '/translation bugs/d'  >>doc/bison.help.tmp
+	      -e '/translation bugs/d'  >>doc/bison.help.tmp
 	$(AM_V_at)$(top_srcdir)/build-aux/move-if-change doc/bison.help.tmp $@
 endif ! CROSS_COMPILING
 
@@ -118,7 +125,7 @@ endif
 $(top_srcdir)/doc/bison.1: $(MAN_DEPS)
 	$(AM_V_GEN)$(HELP2MAN)			\
 	    --include=$(top_srcdir)/doc/bison.x	\
-	    --output=$@.tmp src/bison$(EXEEXT)
+	    --output=$@.tmp tests/bison
 	$(AM_V_at)if $(remove_time_stamp) $@ >$@a.tmp 2>/dev/null &&		\
 	   $(remove_time_stamp) $@.tmp | cmp $@a.tmp - >/dev/null 2>&1; then	\
 	  touch $@;								\
@@ -135,14 +142,14 @@ endif
 ## Graphviz examples generation. ##
 ## ----------------------------- ##
 
-CLEANFILES += $(FIGS_GV:.gv=.eps) $(FIGS_GV:.gv=.pdf) $(FIGS_GV:.gv=.png)
+CLEANFILES += $(FIGS_GV:.gv=.eps) $(FIGS_GV:.gv=.pdf) $(FIGS_GV:.gv=.svg)
 FIGS_GV =                                               \
   doc/figs/example.gv                                   \
   doc/figs/example-reduce.gv doc/figs/example-shift.gv
 EXTRA_DIST +=                                                   \
   $(FIGS_GV) $(FIGS_GV:.gv=.txt)                                \
-  $(FIGS_GV:.gv=.eps) $(FIGS_GV:.gv=.pdf) $(FIGS_GV:.gv=.png)
-SUFFIXES += .gv .eps .pdf .png
+  $(FIGS_GV:.gv=.eps) $(FIGS_GV:.gv=.pdf) $(FIGS_GV:.gv=.svg)
+SUFFIXES += .gv .eps .pdf .svg
 
 .gv.eps:
 	$(AM_V_GEN) $(MKDIR_P) `echo "./$@" | sed -e 's,/[^/]*$$,,'`
@@ -154,9 +161,9 @@ SUFFIXES += .gv .eps .pdf .png
 	$(AM_V_at) $(DOT) -Gmargin=0 -Tpdf $< >$@.tmp
 	$(AM_V_at) mv $@.tmp $@
 
-.gv.png:
+.gv.svg:
 	$(AM_V_GEN) $(MKDIR_P) `echo "./$@" | sed -e 's,/[^/]*$$,,'`
-	$(AM_V_at) $(DOT) -Gmargin=0 -Tpng $< >$@.tmp
+	$(AM_V_at) $(DOT) -Gmargin=0 -Tsvg $< >$@.tmp
 	$(AM_V_at) mv $@.tmp $@
 
 ## -------------- ##

examples/c++/calc++/local.mk

@@ -18,8 +18,6 @@
 ## Parser generation.  ##
 ## ------------------- ##
 
-# Don't depend on $(BISON) otherwise we would rebuild these files
-# in srcdir, including during distcheck, which is forbidden.
 %D%/parser.stamp: $(dependencies)
 SUFFIXES += .yy .stamp
 .yy.stamp:
@@ -33,6 +31,7 @@ $(calcxx_sources_generated): %D%/parser.stamp
 	@test -f $@ || $(MAKE) $(AM_MAKEFLAGS) %D%/parser.stamp
 CLEANFILES +=					\
   $(calcxx_sources_generated)			\
+  %D%/parser.dot				\
   %D%/parser.output				\
   %D%/parser.stamp				\
   %D%/scanner.cc

examples/c++/local.mk

@@ -68,4 +68,6 @@ endif
 EXTRA_DIST += %D%/variant-11.test
 
 dist_cxx_DATA = %D%/README.md %D%/Makefile %D%/variant.yy %D%/variant-11.yy
-CLEANFILES += %D%/simple.output %D%/variant.output %D%/variant-11.output
+CLEANFILES +=                                                   \
+  %D%/simple.output %D%/variant.output %D%/variant-11.output    \
+  %D%/simple.hh %D%/variant.hh %D%/variant-11.hh

examples/c/README.md

@@ -1,9 +1,11 @@
-This directory contains examples of Bison grammar files.
+# Examples in C
 
-Most of them come from the documentation, which should be installed together
+This directory contains simple examples of Bison grammar files in C.
+
+Some of them come from the documentation, which should be installed together
 with Bison.  The URLs are provided for convenience.
 
-# rpcalc - Reverse Polish Notation Calculator
+## rpcalc - Reverse Polish Notation Calculator
 The first example is that of a simple double-precision Reverse Polish
 Notation calculator (a calculator using postfix operators). This example
 provides a good starting point, since operator precedence is not an issue.
@@ -11,12 +13,30 @@ provides a good starting point, since operator precedence is not an issue.
 Extracted from the documentation: "Reverse Polish Notation Calculator"
 https://www.gnu.org/software/bison/manual/html_node/RPN-Calc.html
 
-# mfcalc - Multi-Function Calculator
-A more complete C example: a multi-function calculator.
+## calc - Simple Calculator
+This example is slightly more complex than rpcalc: it features infix
+operators (`1 + 2`, instead of `1 2 +` in rpcalc), but it does so using a
+unambiguous grammar of the arithmetic instead of using precedence
+directives (%left, etc.).
+
+## mfcalc - Multi-Function Calculator
+A more complete C example: a multi-function calculator.  More complex than
+the previous example.  Using precedence directives to support infix
+operators.
 
 Extracted from the documentation: "Multi-Function Calculator: mfcalc".
 https://www.gnu.org/software/bison/manual/html_node/Multi_002dfunction-Calc.html
 
+## lexcalc - calculator with Flex and Bison
+The calculator, redux.  This time using a scanner generated by Flex.
+
+## reccalc - recursive calculator with Flex and Bison
+The example builds on top of the previous one to provide a reentrant parser.
+Such parsers can be called concurrently in different threads, or even
+recursively.  To demonstrate this feature, expressions in parentheses are
+tokenized as strings, and then recursively parsed from the parser.  So
+`(((1)+(2))*((3)+(4)))` uses eight parsers, with a depth of four.
+
 
 <!---
 
@@ -34,5 +54,6 @@ Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
 Texts.  A copy of the license is included in the "GNU Free
 Documentation License" file as part of this distribution.
 
-# LocalWords:  mfcalc calc parsers yy
+# LocalWords:  mfcalc calc parsers yy rpcalc lexcalc redux reccalc ispell
+# LocalWords:  reentrant tokenized american postfix
 --->

examples/c/calc/Makefile

@@ -0,0 +1,28 @@
+# This Makefile is designed to be simple and readable.  It does not
+# aim at portability.  It requires GNU Make.
+
+BASE = calc
+BISON = bison
+XSLTPROC = xsltproc
+
+all: $(BASE)
+
+%.c %.h %.xml %.gv: %.y
+	$(BISON) $(BISONFLAGS) --defines --xml --graph=$*.gv -o $*.c $<
+
+$(BASE): $(BASE).o
+	$(CC) $(CFLAGS) -o $@ $^
+
+run: $(BASE)
+	@echo "Type arithmetic expressions.  Quit with ctrl-d."
+	./$<
+
+html: $(BASE).html
+%.html: %.xml
+	$(XSLTPROC) $(XSLTPROCFLAGS) -o $@ $$($(BISON) --print-datadir)/xslt/xml2xhtml.xsl $<
+
+CLEANFILES =									\
+  $(BASE) *.o $(BASE).[ch] $(BASE).output $(BASE).xml $(BASE).html $(BASE).gv
+
+clean:
+	rm -f $(CLEANFILES)

examples/c/calc/README.md

@@ -0,0 +1,28 @@
+# calc - calculator with Bison
+
+This directory contains calc, the traditional example of using Bison to
+build a simple calculator.
+
+<!---
+Local Variables:
+fill-column: 76
+ispell-dictionary: "american"
+End:
+
+Copyright (C) 2019 Free Software Foundation, Inc.
+
+This file is part of Bison, the GNU Compiler Compiler.
+
+This program is free software: you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program.  If not, see <http://www.gnu.org/licenses/>.
+--->

examples/c/calc/calc.test

@@ -0,0 +1,37 @@
+#! /bin/sh
+
+# Copyright (C) 2019 Free Software Foundation, Inc.
+#
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+cat >input <<EOF
+1+2*3
+EOF
+run 0 7
+
+cat >input <<EOF
+1 - 2 - 3
+EOF
+run 0 -4
+
+cat >input <<EOF
+8 / 2 / 2
+EOF
+run 0 2
+
+cat >input <<EOF
+(1+2) * 3
+EOF
+run 0 9
+run -noerr 0 9 -p

examples/c/calc/calc.y

@@ -0,0 +1,93 @@
+%code top {
+  #include <ctype.h>  /* isdigit. */
+  #include <stdio.h>  /* For printf, etc. */
+  #include <string.h> /* strcmp. */
+
+  int yylex (void);
+  void yyerror (char const *);
+}
+
+%define api.header.include {"calc.h"}
+%define api.value.type union /* Generate YYSTYPE from these types:  */
+%token <double> NUM "number"
+%type  <double> expr term fact
+
+/* Generate the parser description file.  */
+%verbose
+/* Enable run-time traces (yydebug).  */
+%define parse.trace
+
+/* Formatting semantic values.  */
+%printer { fprintf (yyo, "%g", $$); } <double>;
+
+%% /* The grammar follows.  */
+input:
+  %empty
+| input line
+;
+
+line:
+  '\n'
+| expr '\n'  { printf ("%.10g\n", $1); }
+| error '\n' { yyerrok; }
+;
+
+expr:
+  expr '+' term { $$ = $1 + $3; }
+| expr '-' term { $$ = $1 - $3; }
+| term
+;
+
+term:
+  term '*' fact { $$ = $1 * $3; }
+| term '/' fact { $$ = $1 / $3; }
+| fact
+;
+
+fact:
+  "number"
+| '(' expr ')' { $$ = $2; }
+;
+
+%%
+
+int
+yylex (void)
+{
+  int c;
+
+  /* Ignore white space, get first nonwhite character.  */
+  while ((c = getchar ()) == ' ' || c == '\t')
+    continue;
+
+  if (c == EOF)
+    return 0;
+
+  /* Char starts a number => parse the number.         */
+  if (c == '.' || isdigit (c))
+    {
+      ungetc (c, stdin);
+      scanf ("%lf", &yylval.NUM);
+      return NUM;
+    }
+
+  /* Any other character is a token by itself.        */
+  return c;
+}
+
+/* Called by yyparse on error.  */
+void
+yyerror (char const *s)
+{
+  fprintf (stderr, "%s\n", s);
+}
+
+int
+main (int argc, char const* argv[])
+{
+  /* Enable parse traces on option -p.  */
+  for (int i = 1; i < argc; ++i)
+    if (!strcmp (argv[i], "-p"))
+      yydebug = 1;
+  return yyparse ();
+}

examples/c/calc/local.mk

@@ -0,0 +1,33 @@
+## Copyright (C) 2019 Free Software Foundation, Inc.
+##
+## This program is free software: you can redistribute it and/or modify
+## it under the terms of the GNU General Public License as published by
+## the Free Software Foundation, either version 3 of the License, or
+## (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+calcdir = $(docdir)/%D%
+
+## ------ ##
+## Calc.  ##
+## ------ ##
+
+check_PROGRAMS += %D%/calc
+TESTS += %D%/calc.test
+EXTRA_DIST += %D%/calc.test
+nodist_%C%_calc_SOURCES = %D%/calc.y
+%D%/calc.c: $(dependencies)
+
+# Don't use gnulib's system headers.
+%C%_calc_CPPFLAGS = -I$(top_srcdir)/%D% -I$(top_builddir)/%D%
+
+dist_calc_DATA = %D%/calc.y %D%/Makefile %D%/README.md
+CLEANFILES += %D%/calc.[ch] %D%/calc.output %D%/scan.c
+CLEANDIRS += %D%/*.dSYM

examples/c/lexcalc/Makefile

@@ -15,7 +15,7 @@ all: $(BASE)
 	$(FLEX) $(FLEXFLAGS) -o$@ $<
 
 scan.o: parse.h
-lexcalc: parse.o scan.o
+$(BASE): parse.o scan.o
 	$(CC) $(CFLAGS) -o $@ $^
 
 run: $(BASE)

examples/c/lexcalc/lexcalc.test

@@ -25,3 +25,8 @@ cat >input <<EOF
 EOF
 run 0 9
 run -noerr 0 9 -p
+
+cat >input <<EOF
+(1+2) *
+EOF
+run 1 'err: syntax error, unexpected end-of-line, expecting ( or number'

examples/c/lexcalc/local.mk

@@ -22,11 +22,12 @@ lexcalcdir = $(docdir)/%D%
 check_PROGRAMS += %D%/lexcalc
 TESTS += %D%/lexcalc.test
 EXTRA_DIST += %D%/lexcalc.test
-%C%_lexcalc_SOURCES = %D%/parse.y %D%/parse.h %D%/scan.l
+nodist_%C%_lexcalc_SOURCES = %D%/parse.y %D%/parse.h %D%/scan.l
+%D%/parse.c: $(dependencies)
 
 # Don't use gnulib's system headers.
 %C%_lexcalc_CPPFLAGS = -I$(top_srcdir)/%D% -I$(top_builddir)/%D%
 
 dist_lexcalc_DATA = %D%/parse.y %D%/scan.l %D%/Makefile %D%/README.md
-CLEANFILES += %D%/lexcalc %D%/*.o %D%/parse.c %D%/scan.c
+CLEANFILES += %D%/parse.[ch] %D%/scan.c %D%/parse.output
 CLEANDIRS += %D%/*.dSYM

examples/c/lexcalc/parse.y

@@ -15,8 +15,8 @@
 // Emitted on top of the implementation file.
 %code top
 {
-#include <stdio.h>  /* printf. */
-#include <stdlib.h> /* getenv. */
+#include <stdio.h>  // printf.
+#include <stdlib.h> // getenv.
 }
 
 %define api.pure full
@@ -39,7 +39,7 @@
 ;
 
 %token <int> NUM "number"
-%type <int> exp line
+%type <int> exp
 %printer { fprintf (yyo, "%d", $$); } <int>
 
 // Precedence (from lowest to highest) and associativity.
@@ -50,18 +50,18 @@
 // Rules.
 input:
   %empty
-| input line  { printf ("%d\n", $line); }
+| input line
 ;
 
 line:
-  exp EOL   { $$ = $1; }
+  exp EOL   { printf ("%d\n", $exp); }
 | error EOL { yyerrok; }
 ;
 
 exp:
-  exp "+" exp   { $$ = $1 + $3;  }
-| exp "-" exp   { $$ = $1 - $3;  }
-| exp "*" exp   { $$ = $1 * $3;  }
+  exp "+" exp   { $$ = $1 + $3; }
+| exp "-" exp   { $$ = $1 - $3; }
+| exp "*" exp   { $$ = $1 * $3; }
 | exp "/" exp
   {
     if ($3 == 0)
@@ -72,15 +72,15 @@ exp:
     else
       $$ = $1 / $3;
   }
-| "(" exp ")"   { $$ = $2;       }
+| "(" exp ")"   { $$ = $2; }
 | NUM           { $$ = $1; }
 ;
 %%
 // Epilogue (C code).
-void yyerror(int *nerrs, const char *msg)
+void yyerror (int *nerrs, const char *msg)
 {
   fprintf (stderr, "%s\n", msg);
-  ++nerrs;
+  ++*nerrs;
 }
 
 int main (void)

examples/c/local.mk

@@ -16,6 +16,8 @@
 cdir = $(docdir)/%D%
 dist_c_DATA = %D%/README.md
 
+include %D%/calc/local.mk
 include %D%/lexcalc/local.mk
 include %D%/mfcalc/local.mk
+include %D%/reccalc/local.mk
 include %D%/rpcalc/local.mk

examples/c/mfcalc/local.mk

@@ -28,6 +28,7 @@ extracted += $(mfcalc_extracted)
 
 check_PROGRAMS += %D%/mfcalc
 nodist_%C%_mfcalc_SOURCES = $(mfcalc_sources)
+%D%/mfcalc.c: $(dependencies)
 # Don't use gnulib's system headers.
 %C%_mfcalc_CPPFLAGS = -I$(top_srcdir)/%D% -I$(top_builddir)/%D%
 %C%_mfcalc_LDADD = -lm

examples/c/reccalc/Makefile

@@ -0,0 +1,35 @@
+# This Makefile is designed to be simple and readable.  It does not
+# aim at portability.  It requires GNU Make.
+
+BASE = reccalc
+BISON = bison
+FLEX = flex
+XSLTPROC = xsltproc
+
+all: $(BASE)
+
+%.c %.h %.xml %.gv: %.y
+	$(BISON) $(BISONFLAGS) --defines --xml --graph=$*.gv -o $*.c $<
+
+%.c %.h: %.l
+	$(FLEX) $(FLEXFLAGS) -o$*.c --header-file=$*.h $<
+
+scan.o: parse.h
+parse.o: scan.h
+$(BASE): parse.o scan.o
+	$(CC) $(CFLAGS) -o $@ $^
+
+run: $(BASE)
+	@echo "Type arithmetic expressions.  Quit with ctrl-d."
+	./$<
+
+html: parse.html
+%.html: %.xml
+	$(XSLTPROC) $(XSLTPROCFLAGS) -o $@ $$($(BISON) --print-datadir)/xslt/xml2xhtml.xsl $<
+
+CLEANFILES =						\
+  $(BASE) *.o						\
+  parse.[ch] parse.output parse.xml parse.html parse.gv	\
+  scan.[ch]
+clean:
+	rm -f $(CLEANFILES)

examples/c/reccalc/README.md

@@ -0,0 +1,45 @@
+# reccalc - recursive calculator with Flex and Bison
+
+In this example the generated parser is pure and reentrant: it can be used
+concurrently in different threads, or recursively.  As a proof of this
+reentrancy, expressions in parenthesis are tokenized as strings, and then
+recursively parsed from the parser:
+
+```
+exp: STR
+  {
+    result r = parse_string ($1);
+    free ($1);
+    if (r.nerrs)
+      {
+        res->nerrs += r.nerrs;
+        YYERROR;
+      }
+    else
+      $$ = r.value;
+  }
+```
+
+<!---
+Local Variables:
+fill-column: 76
+ispell-dictionary: "american"
+End:
+
+Copyright (C) 2018-2019 Free Software Foundation, Inc.
+
+This file is part of Bison, the GNU Compiler Compiler.
+
+This program is free software: you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program.  If not, see <http://www.gnu.org/licenses/>.
+--->

examples/c/reccalc/local.mk

@@ -0,0 +1,52 @@
+## Copyright (C) 2019 Free Software Foundation, Inc.
+##
+## This program is free software: you can redistribute it and/or modify
+## it under the terms of the GNU General Public License as published by
+## the Free Software Foundation, either version 3 of the License, or
+## (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+reccalcdir = $(docdir)/%D%
+
+## ------ ##
+## Calc.  ##
+## ------ ##
+
+check_PROGRAMS += %D%/reccalc
+TESTS += %D%/reccalc.test
+EXTRA_DIST += %D%/reccalc.test %D%/scan.l
+nodist_%C%_reccalc_SOURCES = %D%/parse.y %D%/scan.h %D%/scan.c
+BUILT_SOURCES += $(nodist_%C%_reccalc_SOURCES)
+%D%/parse.c: $(dependencies)
+
+# Tell Make that parse.o depends on scan.h, so that scan.h is built
+# before parse.o.  Obfuscate the name of the target, otherwise
+# Automake removes its recipe for parse.o and leaves only our
+# additional dependency.
+DASH = -
+%D%/reccalc$(DASH)parse.o: %D%/scan.h
+
+%D%/scan.c %D%/scan.h: %D%/scan.stamp
+	@test -f $@ || rm -f %D%/scan.stamp
+	@test -f $@ || $(MAKE) $(AM_MAKEFLAGS) %D%/scan.stamp
+
+%D%/scan.stamp: %D%/scan.l
+	$(AM_V_LEX)rm -f $@ $@.tmp
+	$(AM_V_at)$(MKDIR_P) %D%
+	$(AM_V_at)touch $@.tmp
+	$(AM_V_at) $(LEX) -o %D%/scan.c --header-file=%D%/scan.h $<
+	$(AM_V_at)mv $@.tmp $@
+
+# Don't use gnulib's system headers.
+%C%_reccalc_CPPFLAGS = -I$(top_srcdir)/%D% -I$(top_builddir)/%D%
+
+dist_reccalc_DATA = %D%/parse.y %D%/scan.l %D%/Makefile %D%/README.md
+CLEANFILES += %D%/parse.[ch] %D%/parse.output %D%/scan.[ch] %D%/*.stamp
+CLEANDIRS += %D%/*.dSYM

examples/c/reccalc/parse.y

@@ -0,0 +1,183 @@
+// Prologue (directives).
+%expect 0
+
+// Emitted in the header file, before the definition of YYSTYPE.
+%code requires
+{
+  typedef void* yyscan_t;
+  typedef struct
+  {
+    // Whether to print the intermediate results.
+    int verbose;
+    // Value of the last computation.
+    int value;
+    // Number of errors.
+    int nerrs;
+  } result;
+}
+
+// Emitted in the header file, after the definition of YYSTYPE.
+%code provides
+{
+  // Tell Flex the expected prototype of yylex.
+  // The scanner argument must be named yyscanner.
+#define YY_DECL                                                         \
+  enum yytokentype yylex (YYSTYPE* yylval, yyscan_t yyscanner, result *res)
+  YY_DECL;
+
+  void yyerror (yyscan_t scanner, result *res, const char *msg, ...);
+}
+
+// Emitted on top of the implementation file.
+%code top
+{
+#include <stdarg.h> // va_list.
+#include <stdio.h>  // printf.
+#include <stdlib.h> // getenv.
+}
+
+%code
+{
+  result parse_string (const char* cp);
+  result parse (void);
+}
+
+%define api.pure full
+%define api.token.prefix {TOK_}
+%define api.value.type union
+%define parse.error verbose
+%define parse.trace
+%verbose
+
+ // Scanner and error count are exchanged between main, yyparse and yylex.
+%param {yyscan_t scanner}{result *res}
+
+%token
+  PLUS   "+"
+  MINUS  "-"
+  STAR   "*"
+  SLASH  "/"
+  EOL    "end-of-line"
+  EOF 0  "end-of-file"
+;
+
+%token <int> NUM "number"
+%type <int> exp
+%printer { fprintf (yyo, "%d", $$); } <int>
+
+%token <char*> STR "string"
+%printer { fprintf (yyo, "\"%s\"", $$); } <char*>
+%destructor { free ($$); } <char*>
+
+// Precedence (from lowest to highest) and associativity.
+%left "+" "-"
+%left "*" "/"
+%precedence UNARY
+
+%%
+// Rules.
+input:
+  line
+| input line
+;
+
+line:
+  exp eol
+    {
+      res->value = $exp;
+      if (res->verbose)
+        printf ("%d\n", $exp);
+    }
+| error eol
+    {
+      yyerrok;
+    }
+;
+
+eol:
+  EOF
+| EOL
+;
+
+exp:
+  NUM           { $$ = $1; }
+| exp "+" exp   { $$ = $1 + $3; }
+| exp "-" exp   { $$ = $1 - $3; }
+| exp "*" exp   { $$ = $1 * $3; }
+| exp "/" exp
+  {
+    if ($3 == 0)
+      {
+        yyerror (scanner, res, "invalid division by zero");
+        YYERROR;
+      }
+    else
+      $$ = $1 / $3;
+  }
+| "+" exp %prec UNARY  { $$ = + $2; }
+| "-" exp %prec UNARY  { $$ = - $2; }
+| STR
+  {
+    result r = parse_string ($1);
+    free ($1);
+    if (r.nerrs)
+      {
+        res->nerrs += r.nerrs;
+        YYERROR;
+      }
+    else
+      $$ = r.value;
+  }
+;
+
+%%
+// Epilogue (C code).
+
+#include "scan.h"
+
+result
+parse (void)
+{
+  yyscan_t scanner;
+  yylex_init (&scanner);
+  result res = {1, 0, 0};
+  yyparse (scanner, &res);
+  yylex_destroy (scanner);
+  return res;
+}
+
+result
+parse_string (const char *str)
+{
+  yyscan_t scanner;
+  yylex_init (&scanner);
+  YY_BUFFER_STATE buf = yy_scan_string (str ? str : "", scanner);
+  result res = {0, 0, 0};
+  yyparse (scanner, &res);
+  yy_delete_buffer (buf, scanner);
+  yylex_destroy (scanner);
+  return res;
+}
+
+void
+yyerror (yyscan_t scanner, result *res,
+         const char *msg, ...)
+{
+  (void) scanner;
+  va_list args;
+  va_start (args, msg);
+  vfprintf (stderr, msg, args);
+  va_end (args);
+  fputc ('\n', stderr);
+  res->nerrs += 1;
+}
+
+int
+main (void)
+{
+  // Possibly enable parser runtime debugging.
+  yydebug = !!getenv ("YYDEBUG");
+  result res = parse ();
+  // Exit on failure if there were errors.
+  return !!res.nerrs;
+}

examples/c/reccalc/reccalc.test

@@ -0,0 +1,57 @@
+#! /bin/sh
+
+# Copyright (C) 2018-2019 Free Software Foundation, Inc.
+#
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+cat >input <<EOF
+1+2*3
+EOF
+run 0 '7'
+
+cat >input <<EOF
+(1+2) * 3
+EOF
+run 0 '9'
+run -noerr 0 '9' -p
+
+cat >input <<EOF
+(((1)+(2))*((3)+(4)))
+EOF
+run 0 '21'
+
+# Some really deep computation.
+# for 4, gives 4 + (3 + (2 + (1 + (- (4 * (4 + 1)) / 2)))).
+n=100
+for i in $(seq 0 $n)
+do
+    if [ "$i" -eq 0 ]; then
+        input="- ($n * ($n + 1)) / 2"
+    else
+        input="$i + ($input)"
+    fi
+done
+echo "$input" > input
+run 0 '0'
+
+cat >input <<EOF
+() + ()
+EOF
+run 1 'err: syntax error, unexpected end-of-file, expecting + or - or number or string'
+
+cat >input <<EOF
+1 + $
+EOF
+run 1 'err: syntax error, invalid character: $
+err: syntax error, unexpected end-of-line, expecting + or - or number or string'

examples/c/reccalc/scan.l

@@ -0,0 +1,85 @@
+/* Prologue (directives).   -*- C++ -*- */
+
+/* Disable Flex features we don't need, to avoid warnings. */
+%option nodefault noinput nounput noyywrap
+
+%option reentrant
+
+%{
+#include <assert.h>
+#include <limits.h> /* INT_MIN */
+#include <stdlib.h> /* strtol */
+
+#include "parse.h"
+%}
+
+%x SC_STRING
+
+%%
+%{
+  int nesting = 0;
+  char *str = NULL;
+  int size = 0;
+  int capacity = 0;
+#define STR_APPEND()                                    \
+  do {                                                  \
+    if (capacity < size + yyleng + 1)                   \
+      {                                                 \
+        do                                              \
+          capacity = capacity ? 2 * capacity : 128;     \
+        while (capacity < size + yyleng + 1);           \
+        str = realloc (str, capacity);                  \
+      }                                                 \
+    memcpy (str + size, yytext, yyleng);                \
+    size += yyleng;                                     \
+    assert (size < capacity);                           \
+  } while (0)
+%}
+
+ // Rules.
+
+"+"      return TOK_PLUS;
+"-"      return TOK_MINUS;
+"*"      return TOK_STAR;
+"/"      return TOK_SLASH;
+
+"("      nesting += 1; BEGIN SC_STRING;
+
+ /* Scan an integer.  */
+[0-9]+   {
+  errno = 0;
+  long n = strtol (yytext, NULL, 10);
+  if (! (INT_MIN <= n && n <= INT_MAX && errno != ERANGE))
+    yyerror (yyscanner, res, "integer is out of range");
+  yylval->TOK_NUM = (int) n;
+  return TOK_NUM;
+}
+
+ /* Ignore white spaces. */
+[ \t]+   continue;
+
+"\n"     return TOK_EOL;
+
+.        yyerror (yyscanner, res, "syntax error, invalid character: %c", yytext[0]);
+
+<SC_STRING>
+{
+  "("+   nesting += yyleng; STR_APPEND ();
+  ")"    {
+    if (!--nesting)
+      {
+        BEGIN INITIAL;
+        if (str)
+          str[size] = 0;
+        yylval->TOK_STR = str;
+        return TOK_STR;
+      }
+    else
+      STR_APPEND ();
+  }
+  [^()]+  STR_APPEND ();
+}
+
+<<EOF>>  return TOK_EOF;
+%%
+/* Epilogue (C code). */

examples/c/rpcalc/local.mk

@@ -28,6 +28,7 @@ extracted += $(rpcalc_extracted)
 
 check_PROGRAMS += %D%/rpcalc
 nodist_%C%_rpcalc_SOURCES = $(rpcalc_sources)
+%D%/rpcalc.c: $(dependencies)
 # Don't use gnulib's system headers.
 %C%_rpcalc_CPPFLAGS = -I$(top_builddir)/%D%
 %C%_rpcalc_LDADD = -lm

examples/d/calc.test

@@ -23,4 +23,4 @@ run 0 7
 cat >input <<EOF
 1 + 2 * * 3
 EOF
-run 0 "err: syntax error, unexpected *, expecting - or ! or ( or number"
+run 1 "err: syntax error, unexpected *, expecting + or - or ( or number"

examples/d/calc.y

@@ -18,17 +18,15 @@
        MINUS  "-"
        STAR   "*"
        SLASH  "/"
-       BANG   "!"
        LPAR   "("
        RPAR   ")"
        EOL    "end of line"
 %token <ival> NUM "number"
 %type  <ival> exp
 
-%nonassoc "="       /* comparison            */
 %left "-" "+"
 %left "*" "/"
-%precedence NEG     /* negation--unary minus */
+%precedence UNARY   /* unary operators */
 
 /* Grammar follows */
 %%
@@ -44,96 +42,86 @@ line:
 ;
 
 exp:
-  NUM                { $$ = $1; }
-| exp "=" exp
-  {
-    if ($1 != $3)
-      yyerror (format ("calc: error: %d != %d", $1, $3));
-  }
-| exp "+" exp        { $$ = $1 + $3; }
-| exp "-" exp        { $$ = $1 - $3; }
-| exp "*" exp        { $$ = $1 * $3; }
-| exp "/" exp        { $$ = $1 / $3; }
-| "-" exp  %prec NEG { $$ = -$2; }
-| "(" exp ")"        { $$ = $2; }
-| "(" error ")"      { $$ = 1111; }
-| "!"                { $$ = 0; return YYERROR; }
-| "-" error          { $$ = 0; return YYERROR; }
+  NUM                  { $$ = $1; }
+| exp "+" exp          { $$ = $1 + $3; }
+| exp "-" exp          { $$ = $1 - $3; }
+| exp "*" exp          { $$ = $1 * $3; }
+| exp "/" exp          { $$ = $1 / $3; }
+| "+" exp  %prec UNARY { $$ = -$2; }
+| "-" exp  %prec UNARY { $$ = -$2; }
+| "(" exp ")"          { $$ = $2; }
 ;
 
-
 %%
-class CalcLexer : Lexer {
+import std.range.primitives;
+import std.stdio;
 
-  this ()
-    {}
+auto calcLexer(R)(R range)
+  if (isInputRange!R && is (ElementType!R : dchar))
+{
+  return new CalcLexer!R(range);
+}
 
-  int
-  get_char ()
-  {
-    import stdc = core.stdc.stdio;
-    return stdc.getc (stdc.stdin);
-  }
+auto calcLexer (File f)
+{
+  import std.algorithm : map, joiner;
+  import std.utf : byDchar;
 
-  void
-  unget_char (int c)
-  {
-    import stdc = core.stdc.stdio;
-    stdc.ungetc (c, stdc.stdin);
-  }
+  return f.byChunk(1024)        // avoid making a syscall roundtrip per char
+          .map!(chunk => cast(char[]) chunk) // because byChunk returns ubyte[]
+          .joiner               // combine chunks into a single virtual range of char
+          .calcLexer;           // forward to other overload
+}
+
+class CalcLexer(R) : Lexer
+  if (isInputRange!R && is (ElementType!R : dchar))
+{
+  R input;
+
+  this(R r) { input = r; }
+
+  // Should be a local in main, shared with %parse-param.
+  int exit_status = 0;
 
   public void yyerror (string s)
   {
+    exit_status = 1;
     stderr.writeln (s);
   }
 
-  int
-  read_signed_integer ()
-  {
-    int c = get_char ();
-    int sign = 1;
-    int n = 0;
-
-    if (c == '-')
-      {
-        c = get_char ();
-        sign = -1;
-      }
-
-    while (isDigit (c))
-      {
-        n = 10 * n + (c - '0');
-        c = get_char ();
-      }
-
-    unget_char (c);
-    return sign * n;
-  }
-
   YYSemanticType semanticVal_;
 
-  public final @property YYSemanticType semanticVal()
+  public final @property YYSemanticType semanticVal ()
   {
     return semanticVal_;
   }
 
-  YYTokenType yylex ()
+  int yylex ()
   {
-    int c;
-    /* Skip white spaces.  */
-    do
-      {}
-    while ((c = get_char ()) == ' ' || c == '\t');
+    import std.uni : isWhite, isNumber;
 
-    /* process numbers   */
-    if (c == '.' || isDigit (c))
+    // Skip initial spaces
+    while (!input.empty && input.front != '\n' && isWhite (input.front))
+    {
+      input.popFront;
+    }
+
+    // Handle EOF.
+    if (input.empty)
+      return YYTokenType.EOF;
+
+    // Numbers.
+    if (input.front == '.' || input.front.isNumber)
       {
-        unget_char (c);
-        semanticVal_.ival = read_signed_integer ();
+        import std.conv : parse;
+        semanticVal_.ival = input.parse!int;
         return YYTokenType.NUM;
       }
 
-    switch (c)
+    // Individual characters
+    auto ch = input.front;
+    input.popFront;
+    switch (ch)
       {
       case EOF: return YYTokenType.EOF;
       case '=': return YYTokenType.EQ;
@@ -141,7 +129,6 @@ class CalcLexer : Lexer {
       case '-': return YYTokenType.MINUS;
       case '*': return YYTokenType.STAR;
       case '/': return YYTokenType.SLASH;
-      case '!': return YYTokenType.BANG;
       case '(': return YYTokenType.LPAR;
       case ')': return YYTokenType.RPAR;
       case '\n': return YYTokenType.EOL;
@@ -150,9 +137,10 @@ class CalcLexer : Lexer {
   }
 }
 
-void main ()
+int main ()
 {
-  CalcLexer l = new CalcLexer ();
-  Calc p = new Calc (l);
+  auto l = calcLexer (stdin);
+  auto p = new Calc (l);
   p.parse ();
+  return l.exit_status;
 }

examples/extexi

@@ -60,7 +60,11 @@ sub normalize($)
 {
   local ($_) = @_;
 
-  s/^\@(c |comment|dots|end (ignore|group)|ignore|group).*\n//mg;
+  # If we just remove this lines, then the compiler's tracking of
+  # #lines is broken.  Leave lines that that accepted by all our tools
+  # (including flex, hence the leading space), and that will be easy
+  # to remove (see the Make examples-unline recipe).
+  s{^\@(c |comment|dots|end (ignore|group)|ignore|group).*}{ /**/}mg;
   s/\@value\{VERSION\}/$ENV{VERSION}/g;
   s/^\@(error|result)\{\}//mg;
   s/\@([{}@])/$1/g;

examples/local.mk

@@ -14,6 +14,24 @@
 ## You should have received a copy of the GNU General Public License
 ## along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
+
+## Because some of our examples use
+##
+##     %C%_reccalc_SOURCES = %D%/parse.y
+##
+## Automake ships parse.y and parse.c, and possibly parse.h when it
+## "understands" that there is one.  This is not what we want: ship only
+## parser.y.  Yet we still want to use Automake to compile the sources
+## from parser.y.  The easiest seems to use
+##
+##     nodist_%C%_reccalc_SOURCES = %D%/parse.y
+##
+## together with
+##
+##     dist_reccalc_DATA = %D%/parse.y %D%/scan.l %D%/Makefile %D%/README.md
+##
+## which guarantees that parse.y is indeed shipped.
+
 dist_noinst_SCRIPTS = %D%/extexi %D%/test
 TEST_LOG_COMPILER = $(top_srcdir)/%D%/test
 
@@ -59,12 +77,8 @@ EXTRA_DIST += %D%/extracted.stamp
 dist-hook: examples-unline
 .PHONY: examples-unline
 examples-unline:
-	for e in $(extracted);			\
-	do					\
-	  e=$(distdir)/$$e;			\
-	  sed -e '/#line/d' $$e >$$e.tmp;	\
-	  mv -f $$e.tmp $$e;			\
-	done
+	cd $(distdir) && \
+	  perl -pi -0777 -e 's/#line.*\n//g;s{^ /\*\*/\n}{}mg' $(extracted)
 
 
 ## ---------- ##

gnulib-po/.gitignore

@@ -0,0 +1,15 @@
+/*.gmo
+/*.po
+/LINGUAS
+/Makefile.in.in
+/Makevars
+/Makevars.template
+/POTFILES.in
+/Rules-quot
+/bison-gnulib.pot
+/boldquot.sed
+/[email protected]
+/[email protected]
+/insert-header.sin
+/quot.sed
+/remove-potcdate.sin

lib/.gitignore

@@ -318,3 +318,6 @@
 /rename.c
 /rmdir.c
 /same-inode.h
+/assure.h
+/fsync.c
+/textstyle.in.h

m4/.gitignore

@@ -5,6 +5,8 @@
 /asm-underscore.m4
 /assert.m4
 /calloc.m4
+/canonicalize.m4
+/clock_time.m4
 /close-stream.m4
 /close.m4
 /closeout.m4
@@ -36,16 +38,22 @@
 /frexpl.m4
 /fseterr.m4
 /fstat.m4
+/fsync.m4
 /getdtablesize.m4
+/gethrxtime.m4
 /getopt.m4
 /getprogname.m4
+/getrusage.m4
 /gettext.m4
+/gettime.m4
+/gettimeofday.m4
 /glibc2.m4
 /glibc21.m4
 /gnulib-cache.m4
 /gnulib-common.m4
 /gnulib-comp.m4
 /gnulib-tool.m4
+/host-cpu-c-abi.m4
 /iconv.m4
 /include_next.m4
 /intdiv0.m4
@@ -70,12 +78,15 @@
 /lib-ld.m4
 /lib-link.m4
 /lib-prefix.m4
+/libtextstyle-optional.m4
+/libtextstyle.m4
 /libunistring-base.m4
 /limits-h.m4
 /localcharset.m4
 /locale-fr.m4
 /locale-ja.m4
 /locale-zh.m4
+/localtime-buffer.m4
 /lock.m4
 /longlong.m4
 /lstat.m4
@@ -117,9 +128,15 @@
 /quotearg.m4
 /raise.m4
 /rawmemchr.m4
+/readlink.m4
 /realloc.m4
+/relocatable-lib.m4
+/relocatable.m4
+/rename.m4
+/rmdir.m4
 /sched_h.m4
 /setenv.m4
+/sh-filename.m4
 /sig_atomic_t.m4
 /sigaction.m4
 /signal_h.m4
@@ -132,7 +149,9 @@
 /spawn_h.m4
 /sprintf-posix.m4
 /ssize_t.m4
+/stat-time.m4
 /stat.m4
+/std-gnu11.m4
 /stdbool.m4
 /stddef_h.m4
 /stdint.m4
@@ -148,12 +167,16 @@
 /strndup.m4
 /strnlen.m4
 /strverscmp.m4
+/sys_resource_h.m4
 /sys_socket_h.m4
 /sys_stat_h.m4
+/sys_time_h.m4
+/sys_times_h.m4
 /sys_types_h.m4
 /sys_wait_h.m4
 /threadlib.m4
 /time_h.m4
+/timespec.m4
 /uintmax_t.m4
 /unistd-safer.m4
 /unistd_h.m4
@@ -177,21 +200,3 @@
 /xalloc.m4
 /xsize.m4
 /xstrndup.m4
-/host-cpu-c-abi.m4
-/stat-time.m4
-/sys_time_h.m4
-/sys_times_h.m4
-/sh-filename.m4
-/clock_time.m4
-/gethrxtime.m4
-/getrusage.m4
-/gettime.m4
-/gettimeofday.m4
-/localtime-buffer.m4
-/sys_resource_h.m4
-/timespec.m4
-/intl-thread-locale.m4
-/canonicalize.m4
-/readlink.m4
-/relocatable-lib.m4
-/relocatable.m4

po/POTFILES.in

@@ -9,23 +9,11 @@ src/location.c
 src/main.c
 src/muscle-tab.c
 src/parse-gram.y
+src/print-graph.c
 src/print.c
-src/print_graph.c
 src/reader.c
 src/reduce.c
 src/scan-code.l
 src/scan-gram.l
 src/scan-skel.l
 src/symtab.c
-
-lib/argmatch.c
-lib/closeout.c
-lib/error.c
-lib/getopt.c
-lib/obstack.c
-lib/spawn-pipe.c
-lib/quotearg.c
-lib/timevar.c
-lib/w32spawn.h
-lib/wait-process.c
-lib/xalloc-die.c

src/assoc.c

@@ -29,9 +29,6 @@ assoc_to_string (assoc a)
 {
   switch (a)
     {
-    default:
-      abort ();
-
     case undef_assoc:
       return "undefined associativity";
 
@@ -47,4 +44,5 @@ assoc_to_string (assoc a)
     case precedence_assoc:
       return "%precedence";
     }
+  abort ();
 }

src/closure.c

@@ -35,6 +35,9 @@
 item_number *itemset;
 size_t nitemset;
 
+/* RULESET contains a bit for each rule.  CLOSURE sets the bits for
+   all rules which could potentially describe the next input to be
+   read.  */
 static bitset ruleset;
 
 /* internal data.  See comments before set_fderives and set_firsts.  */
@@ -51,7 +54,7 @@ static bitsetv firsts = NULL;
 `-----------------*/
 
 static void
-print_closure (char const *title, item_number const *array, size_t size)
+closure_print (char const *title, item_number const *array, size_t size)
 {
   fprintf (stderr, "Closure: %s\n", title);
   for (size_t i = 0; i < size; ++i)
@@ -167,7 +170,7 @@ set_fderives (void)
 
 
 void
-new_closure (unsigned n)
+closure_new (unsigned n)
 {
   itemset = xnmalloc (n, sizeof *itemset);
 
@@ -181,8 +184,8 @@ new_closure (unsigned n)
 void
 closure (item_number const *core, size_t n)
 {
-  if (trace_flag & trace_sets)
-    print_closure ("input", core, n);
+  if (trace_flag & trace_closure)
+    closure_print ("input", core, n);
 
   bitset_zero (ruleset);
 
@@ -192,12 +195,12 @@ closure (item_number const *core, size_t n)
 
   /* core is sorted on item index in ritem, which is sorted on rule number.
      Compute itemset with the same sort.  */
+  nitemset = 0;
+  size_t c = 0;
+
   /* A bit index over RULESET. */
   rule_number ruleno;
   bitset_iterator iter;
-
-  nitemset = 0;
-  size_t c = 0;
   BITSET_FOR_EACH (iter, ruleset, ruleno, 0)
     {
       item_number itemno = rules[ruleno].rhs - ritem;
@@ -218,13 +221,13 @@ closure (item_number const *core, size_t n)
       c++;
     }
 
-  if (trace_flag & trace_sets)
-    print_closure ("output", itemset, nitemset);
+  if (trace_flag & trace_closure)
+    closure_print ("output", itemset, nitemset);
 }
 
 
 void
-free_closure (void)
+closure_free (void)
 {
   free (itemset);
   bitset_free (ruleset);

src/closure.h

@@ -27,29 +27,26 @@
    data so that closure can be called.  n is the number of elements to
    allocate for itemset.  */
 
-void new_closure (unsigned n);
+void closure_new (unsigned n);
 
 
 /* Given the kernel (aka core) of a state (a sorted vector of item numbers
    ITEMS, of length N), set up RULESET and ITEMSET to indicate what
    rules could be run and which items could be accepted when those
-   items are the active ones.
-
-   RULESET contains a bit for each rule.  CLOSURE sets the bits for
-   all rules which could potentially describe the next input to be
-   read.
-
-   ITEMSET is a sorted vector of item numbers; NITEMSET is its size
-   (actually, points to just beyond the end of the part of it that is
-   significant).  CLOSURE places there the indices of all items which
-   represent units of input that could arrive next.  */
+   items are the active ones.  */
 
 void closure (item_number const *items, size_t n);
 
 
-/* Frees ITEMSET, RULESET and internal data.  */
+/* Free ITEMSET, RULESET and internal data.  */
 
-void free_closure (void);
+void closure_free (void);
+
+
+/* ITEMSET is a sorted vector of item numbers; NITEMSET is its size
+   (actually, points to just beyond the end of the part of it that is
+   significant).  CLOSURE places there the indices of all items which
+   represent units of input that could arrive next.  */
 
 extern item_number *itemset;
 extern size_t nitemset;

src/complain.c

@@ -23,8 +23,10 @@
 #include "system.h"
 
 #include <argmatch.h>
-#include <stdarg.h>
 #include <progname.h>
+#include <stdarg.h>
+#include <sys/stat.h>
+#include <textstyle.h>
 
 #include "complain.h"
 #include "files.h"
@@ -61,7 +63,47 @@ typedef enum
 /** For each warning type, its severity.  */
 static severity warnings_flag[warnings_size];
 
-static unsigned *indent_ptr = 0;
+static unsigned *indent_ptr = NULL;
+
+styled_ostream_t errstream = NULL;
+
+void
+begin_use_class (const char *s, FILE *out)
+{
+  if (out == stderr)
+    {
+      if (color_debug)
+        fprintf (out, "<%s>", s);
+      else
+        {
+          styled_ostream_begin_use_class (errstream, s);
+          styled_ostream_flush_to_current_style (errstream);
+        }
+    }
+}
+
+void
+end_use_class (const char *s, FILE *out)
+{
+  if (out == stderr)
+    {
+      if (color_debug)
+        fprintf (out, "</%s>", s);
+      else
+        {
+          styled_ostream_end_use_class (errstream, s);
+          styled_ostream_flush_to_current_style (errstream);
+        }
+    }
+}
+
+void
+flush (FILE *out)
+{
+  if (out == stderr)
+    ostream_flush (errstream, FLUSH_THIS_STREAM);
+  fflush (out);
+}
 
 /*------------------------.
 | --warnings's handling.  |
@@ -165,11 +207,89 @@ warnings_argmatch (char *args)
     warning_argmatch ("all", 0, 0);
 }
 
+static const char*
+severity_style (severity s)
+{
+  switch (s)
+    {
+    case severity_disabled:
+    case severity_unset:
+      return "note";
+    case severity_warning:
+      return "warning";
+    case severity_error:
+    case severity_fatal:
+      return "error";
+    }
+  abort ();
+}
+
+static const char*
+severity_prefix (severity s)
+{
+  switch (s)
+    {
+    case severity_disabled:
+    case severity_unset:
+      return "";
+    case severity_warning:
+      return _("warning");
+    case severity_error:
+      return  _("error");
+    case severity_fatal:
+      return _("fatal error");
+    }
+  abort ();
+}
+
 
 /*-----------.
 | complain.  |
 `-----------*/
 
+void
+complain_init_color (void)
+{
+#if HAVE_LIBTEXTSTYLE
+  if (color_mode == color_yes
+      || color_mode == color_html
+      || (color_mode == color_tty && isatty (STDERR_FILENO)))
+    {
+      style_file_prepare ("BISON_DIAGNOSTICS_STYLE", NULL,
+                          pkgdatadir (),
+                          "diagnostics.css");
+      /* As a fallback, use the default in the current directory.  */
+      struct stat statbuf;
+      if ((style_file_name == NULL || stat (style_file_name, &statbuf) < 0)
+          && stat ("diagnostics.css", &statbuf) == 0)
+        style_file_name = "diagnostics.css";
+    }
+  else
+    /* No styling.  */
+    style_file_name = NULL;
+#endif
+
+  /* Workaround clang's warning (starting at Clang 3.5) about the stub
+     code of html_styled_ostream_create:
+
+     | src/complain.c:274:7: error: code will never be executed [-Werror,-Wunreachable-code]
+     |     ? html_styled_ostream_create (file_ostream_create (stderr),
+     |       ^~~~~~~~~~~~~~~~~~~~~~~~~~ */
+#if defined __clang__
+# pragma clang diagnostic push
+# pragma clang diagnostic ignored "-Wunreachable-code"
+#endif
+  errstream =
+    color_mode == color_html
+    ? html_styled_ostream_create (file_ostream_create (stderr),
+                                  style_file_name)
+    : styled_ostream_create (STDERR_FILENO, "(stderr)", TTYCTL_AUTO,
+                             style_file_name);
+#if defined __clang__
+# pragma clang diagnostic pop
+#endif
+}
+
 void
 complain_init (void)
 {
@@ -185,6 +305,12 @@ complain_init (void)
     }
 }
 
+void
+complain_free (void)
+{
+  caret_free ();
+  styled_ostream_free (errstream);
+}
 
 /* A diagnostic with FLAGS is about to be issued.  With what severity?
    (severity_fatal, severity_error, severity_disabled, or
@@ -229,19 +355,30 @@ warning_is_unset (warnings flags)
   return true;
 }
 
-/** Display a "[-Wyacc]" like message on \a f.  */
+bool
+warning_is_enabled (warnings flags)
+{
+  return severity_warning <= warning_severity (flags);
+}
+
+/** Display a "[-Wyacc]" like message on \a out.  */
 
 static void
-warnings_print_categories (warnings warn_flags, FILE *f)
+warnings_print_categories (warnings warn_flags, FILE *out)
 {
-  /* Display only the first match, the second is "-Wall".  */
   for (size_t i = 0; warnings_args[i]; ++i)
     if (warn_flags & warnings_types[i])
       {
         severity s = warning_severity (warnings_types[i]);
-        fprintf (f, " [-W%s%s]",
+        const char* style = severity_style (s);
+        fputs (" [", out);
+        begin_use_class (style, out);
+        fprintf (out, "-W%s%s",
                  s == severity_error ? "error=" : "",
                  warnings_args[i]);
+        end_use_class (style, out);
+        fputc (']', out);
+        /* Display only the first match, the second is "-Wall".  */
         return;
       }
 }
@@ -251,7 +388,8 @@ warnings_print_categories (warnings warn_flags, FILE *f)
  * \param loc     the location, defaulting to the current file,
  *                or the program name.
  * \param flags   the category for this message.
- * \param prefix  put before the message (e.g., "warning").
+ * \param sever   to decide the prefix to put before the message
+ *                (e.g., "warning").
  * \param message the error message, a printf format string.  Iff it
  *                ends with ": ", then no trailing newline is printed,
  *                and the caller should print the remaining
@@ -260,7 +398,7 @@ warnings_print_categories (warnings warn_flags, FILE *f)
  */
 static
 void
-error_message (const location *loc, warnings flags, const char *prefix,
+error_message (const location *loc, warnings flags, severity sever,
                const char *message, va_list args)
 {
   unsigned pos = 0;
@@ -274,7 +412,7 @@ error_message (const location *loc, warnings flags, const char *prefix,
   if (indent_ptr)
     {
       if (*indent_ptr)
-        prefix = NULL;
+        sever = severity_disabled;
       if (!*indent_ptr)
         *indent_ptr = pos;
       else if (*indent_ptr > pos)
@@ -282,13 +420,20 @@ error_message (const location *loc, warnings flags, const char *prefix,
       indent_ptr = NULL;
     }
 
-  if (prefix)
-    fprintf (stderr, "%s: ", prefix);
+  const char* style = severity_style (sever);
+
+  if (sever != severity_disabled)
+    {
+      begin_use_class (style, stderr);
+      fprintf (stderr, "%s:", severity_prefix (sever));
+      end_use_class (style, stderr);
+      fputc (' ', stderr);
+    }
 
   vfprintf (stderr, message, args);
   /* Print the type of warning, only if this is not a sub message
      (in which case the prefix is null).  */
-  if (! (flags & silent) && prefix)
+  if (! (flags & silent) && sever != severity_disabled)
     warnings_print_categories (flags, stderr);
 
   {
@@ -296,16 +441,15 @@ error_message (const location *loc, warnings flags, const char *prefix,
     if (l < 2 || message[l - 2] != ':' || message[l - 1] != ' ')
       {
         putc ('\n', stderr);
-        fflush (stderr);
+        flush (stderr);
         if (loc && feature_flag & feature_caret && !(flags & no_caret))
-          location_caret (*loc, stderr);
+          location_caret (*loc, style, stderr);
       }
   }
-  fflush (stderr);
+  flush (stderr);
 }
 
-/** Raise a complaint. That can be a fatal error, an error or just a
-    warning.  */
+/** Raise a complaint (fatal error, error or just warning).  */
 
 static void
 complains (const location *loc, warnings flags, const char *message,
@@ -317,13 +461,9 @@ complains (const location *loc, warnings flags, const char *message,
 
   if (severity_warning <= s)
     {
-      const char* prefix =
-        s == severity_fatal ? _("fatal error")
-        : s == severity_error ? _("error")
-        : _("warning");
       if (severity_error <= s && ! complaint_status)
         complaint_status = status_warning_as_error;
-      error_message (loc, flags, prefix, message, args);
+      error_message (loc, flags, s, message, args);
     }
 
   if (flags & fatal)
@@ -398,7 +538,7 @@ deprecated_directive (location const *loc, char const *old, char const *upd)
               _("deprecated directive: %s, use %s"),
               quote (old), quote_n (1, upd));
   /* Register updates only if -Wdeprecated is enabled.  */
-  if (warnings_flag[warning_deprecated] != severity_disabled)
+  if (warning_is_enabled (Wdeprecated))
     fixits_register (loc, upd);
 }
 

src/complain.h

@@ -24,6 +24,20 @@
 /* Sub-messages indent. */
 # define SUB_INDENT (4)
 
+/*---------------.
+| Error stream.  |
+`---------------*/
+
+/** Enable a style on \a out provided it's stderr.  */
+void begin_use_class (const char *style, FILE *out);
+
+/** Disable a style on \a out provided it's stderr.  */
+void end_use_class (const char *style, FILE *out);
+
+/** Flush \a out.  */
+void flush (FILE *out);
+
+
 /*-------------.
 | --warnings.  |
 `-------------*/
@@ -76,6 +90,12 @@ void warnings_argmatch (char *args);
 /** Initialize this module.  */
 void complain_init (void);
 
+/** Reclaim resources.  */
+void complain_free (void);
+
+/** Initialize support for colored messages.  */
+void complain_init_color (void);
+
 typedef enum
   {
     Wnone             = 0,       /**< Issue no warnings.  */
@@ -105,6 +125,9 @@ typedef enum
     (Never enabled, never disabled). */
 bool warning_is_unset (warnings flags);
 
+/** Whether warnings of \a flags should be reported. */
+bool warning_is_enabled (warnings flags);
+
 /** Make a complaint, with maybe a location.  */
 void complain (location const *loc, warnings flags, char const *message, ...)
   __attribute__ ((__format__ (__printf__, 3, 4)));

src/conflicts.c

@@ -23,13 +23,13 @@
 
 #include <bitset.h>
 
-#include "LR0.h"
 #include "complain.h"
 #include "conflicts.h"
 #include "files.h"
 #include "getargs.h"
 #include "gram.h"
 #include "lalr.h"
+#include "lr0.h"
 #include "print-xml.h"
 #include "reader.h"
 #include "state.h"
@@ -369,13 +369,16 @@ set_conflicts (state *s, symbol **errors)
     s->solved_conflicts_xml = obstack_finish0 (&solved_conflicts_xml_obstack);
 
   /* Loop over all rules which require lookahead in this state.  Check
-     for conflicts not resolved above.  */
-  for (int i = 0; i < reds->num; ++i)
-    {
-      if (!bitset_disjoint_p (reds->lookahead_tokens[i], lookahead_set))
-        conflicts[s->number] = 1;
-      bitset_or (lookahead_set, lookahead_set, reds->lookahead_tokens[i]);
-    }
+     for conflicts not resolved above.
+
+     reds->lookahead_tokens can be NULL if the LR type is LR(0).  */
+  if (reds->lookahead_tokens)
+    for (int i = 0; i < reds->num; ++i)
+      {
+        if (!bitset_disjoint_p (reds->lookahead_tokens[i], lookahead_set))
+          conflicts[s->number] = 1;
+        bitset_or (lookahead_set, lookahead_set, reds->lookahead_tokens[i]);
+      }
 }
 
 

src/derives.c

@@ -46,9 +46,8 @@ print_derives (void)
 
   for (i = ntokens; i < nsyms; i++)
     {
-      rule **rp;
       fprintf (stderr, "  %s derives\n", symbols[i]->tag);
-      for (rp = derives[i - ntokens]; *rp; ++rp)
+      for (rule **rp = derives[i - ntokens]; *rp; ++rp)
         {
           fprintf (stderr, "    %3d ", (*rp)->user_number);
           rule_rhs_print (*rp, stderr);
@@ -63,10 +62,6 @@ print_derives (void)
 void
 derives_compute (void)
 {
-  symbol_number i;
-  rule_number r;
-  rule **q;
-
   /* DSET[NTERM - NTOKENS] -- A linked list of the numbers of the rules
      whose LHS is NTERM.  */
   rule_list **dset = xcalloc (nvars, sizeof *dset);
@@ -76,7 +71,7 @@ derives_compute (void)
      indexed by rule numbers.  */
   rule_list *delts = xnmalloc (nrules, sizeof *delts);
 
-  for (r = nrules - 1; r >= 0; --r)
+  for (rule_number r = nrules - 1; r >= 0; --r)
     {
       symbol_number lhs = rules[r].lhs->number;
       rule_list *p = &delts[r];
@@ -90,9 +85,10 @@ derives_compute (void)
      it a single array.  */
 
   derives = xnmalloc (nvars, sizeof *derives);
-  q = xnmalloc (nvars + nrules, sizeof *q);
+  /* Q is the storage for DERIVES[...] (DERIVES[0] = q).  */
+  rule **q = xnmalloc (nvars + nrules, sizeof *q);
 
-  for (i = ntokens; i < nsyms; i++)
+  for (symbol_number i = ntokens; i < nsyms; i++)
     {
       rule_list *p = dset[i - ntokens];
       derives[i - ntokens] = q;

src/files.c

@@ -21,11 +21,13 @@
 #include <config.h>
 #include "system.h"
 
+#include <configmake.h> /* PKGDATADIR */
 #include <error.h>
 #include <dirname.h>
 #include <get-errno.h>
 #include <quote.h>
 #include <quotearg.h>
+#include <relocatable.h> /* relocate2 */
 #include <stdio-safer.h>
 #include <xstrndup.h>
 
@@ -49,7 +51,7 @@ location spec_name_prefix_loc = EMPTY_LOCATION_INIT;
 char *spec_verbose_file = NULL;  /* for --verbose. */
 char *spec_graph_file = NULL;    /* for -g. */
 char *spec_xml_file = NULL;      /* for -x. */
-char *spec_defines_file = NULL;  /* for --defines. */
+char *spec_header_file = NULL;  /* for --defines. */
 char *parser_file_name;
 
 /* All computed output file names.  */
@@ -333,14 +335,15 @@ compute_output_file_names (void)
 
   if (defines_flag)
     {
-      if (! spec_defines_file)
-        spec_defines_file = concat2 (all_but_ext, header_extension);
+      if (! spec_header_file)
+        spec_header_file = concat2 (all_but_ext, header_extension);
     }
 
   if (graph_flag)
     {
       if (! spec_graph_file)
-        spec_graph_file = concat2 (all_but_tab_ext, ".dot");
+        spec_graph_file = concat2 (all_but_tab_ext,
+                                   304 <= required_version ? ".gv" : ".dot");
       output_file_name_check (&spec_graph_file, false);
     }
 
@@ -404,6 +407,21 @@ unlink_generated_sources (void)
       unlink (generated_files[i].name);
 }
 
+/* Memory allocated by relocate2, to free.  */
+static char *relocate_buffer = NULL;
+
+char const *
+pkgdatadir (void)
+{
+  if (relocate_buffer)
+    return relocate_buffer;
+  else
+    {
+      char const *cp = getenv ("BISON_PKGDATADIR");
+      return cp ? cp : relocate2 (PKGDATADIR, &relocate_buffer);
+    }
+}
+
 void
 output_file_names_free (void)
 {
@@ -411,10 +429,11 @@ output_file_names_free (void)
   free (spec_verbose_file);
   free (spec_graph_file);
   free (spec_xml_file);
-  free (spec_defines_file);
+  free (spec_header_file);
   free (parser_file_name);
   free (dir_prefix);
   for (int i = 0; i < generated_files_size; i++)
     free (generated_files[i].name);
   free (generated_files);
+  free (relocate_buffer);
 }

src/files.h

@@ -48,7 +48,7 @@ extern char *spec_graph_file;
 extern char *spec_xml_file;
 
 /* File name specified with --defines.  */
-extern char *spec_defines_file;
+extern char *spec_header_file;
 
 /* Directory prefix of output file names.  */
 extern char *dir_prefix;
@@ -64,6 +64,9 @@ extern uniqstr current_file;
 /* The computed base for output file names.  */
 extern char *all_but_ext;
 
+/* Where our data files are installed.  */
+char const *pkgdatadir (void);
+
 void compute_output_file_names (void);
 void output_file_names_free (void);
 

src/fixits.c

@@ -52,6 +52,11 @@ fixit_new (location const *loc, char const* fix)
   return res;
 }
 
+static int
+fixit_cmp (const  fixit *a, const fixit *b)
+{
+  return location_cmp (a->location, b->location);
+}
 
 static void
 fixit_free (fixit *f)
@@ -69,8 +74,8 @@ fixit_print (fixit const *f, FILE *out)
 {
   fprintf (out, "fix-it:%s:{%d:%d-%d:%d}:%s\n",
            quotearg_n_style (1, c_quoting_style, f->location.start.file),
-           f->location.start.line, f->location.start.column,
-           f->location.end.line, f->location.end.column,
+           f->location.start.line, f->location.start.byte,
+           f->location.end.line, f->location.end.byte,
            quotearg_n_style (2, c_quoting_style, f->fix));
 }
 
@@ -85,7 +90,7 @@ fixits_register (location const *loc, char const* fix)
                                    (gl_listelement_dispose_fn) fixit_free,
                                    true);
   fixit *f = fixit_new (loc, fix);
-  gl_list_add_last (fixits, f);
+  gl_sortedlist_add (fixits, (gl_listelement_compar_fn) fixit_cmp, f);
   if (feature_flag & feature_fixit_parsable)
     fixit_print (f, stderr);
 }
@@ -136,19 +141,29 @@ fixits_run (void)
             }
           putc (c, out);
         }
+
       /* Look for the right offset. */
-      while (offset < f->location.start.column)
+      bool need_eol = false;
+      while (offset < f->location.start.byte)
         {
           int c = getc (in);
           if (c == EOF)
             break;
           ++offset;
-          putc (c, out);
+          if (c == '\n')
+            /* The position we are asked for is beyond the actual
+               line: pad with spaces, and remember we need a \n.  */
+            need_eol = true;
+          putc (need_eol ? ' ' : c, out);
         }
 
       /* Paste the fix instead. */
       fputs (f->fix, out);
 
+      /* Maybe install the eol afterwards.  */
+      if (need_eol)
+        putc ('\n', out);
+
       /* Skip the bad input. */
       while (line < f->location.end.line)
         {
@@ -161,16 +176,17 @@ fixits_run (void)
               offset = 1;
             }
         }
-      while (offset < f->location.end.column)
+      while (offset < f->location.end.byte)
         {
           int c = getc (in);
           if (c == EOF)
             break;
           ++offset;
         }
+
       /* If erasing the content of a full line, also remove the
          end-of-line. */
-      if (f->fix[0] == 0 && f->location.start.column == 1)
+      if (f->fix[0] == 0 && f->location.start.byte == 1)
         {
           int c = getc (in);
           if (c == EOF)

src/getargs.c

@@ -19,8 +19,9 @@
    along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
 
 #include <config.h>
+#include "getargs.h"
+
 #include "system.h"
-#include "output.h"
 
 #include <argmatch.h>
 #include <c-strcase.h>
@@ -28,12 +29,13 @@
 #include <error.h>
 #include <getopt.h>
 #include <progname.h>
+#include <quote.h>
+#include <textstyle.h>
 
 #include "complain.h"
 #include "files.h"
-#include "getargs.h"
 #include "muscle-tab.h"
-#include "quote.h"
+#include "output.h"
 #include "uniqstr.h"
 
 bool defines_flag = false;
@@ -43,6 +45,7 @@ bool no_lines_flag = false;
 bool token_table_flag = false;
 location yacc_loc = EMPTY_LOCATION_INIT;
 bool update_flag = false; /* for -u */
+bool color_debug = false;
 
 bool nondeterministic_parser = false;
 bool glr_parser = false;
@@ -182,10 +185,12 @@ ARGMATCH_VERIFY (report_args, report_types);
 static const char * const trace_args[] =
 {
   "none       - no traces",
+  "locations  - full display of the locations",
   "scan       - grammar scanner traces",
   "parse      - grammar parser traces",
   "automaton  - construction of the automaton",
   "bitsets    - use of bitsets",
+  "closure    - input/output of closure",
   "grammar    - reading, reducing the grammar",
   "resource   - memory consumption (where available)",
   "sets       - grammar sets: firsts, nullable etc.",
@@ -202,10 +207,12 @@ static const char * const trace_args[] =
 static const int trace_types[] =
 {
   trace_none,
+  trace_locations,
   trace_scan,
   trace_parse,
   trace_automaton,
   trace_bitsets,
+  trace_closure,
   trace_grammar,
   trace_resource,
   trace_sets,
@@ -230,6 +237,7 @@ static const char * const feature_args[] =
   "none",
   "caret", "diagnostics-show-caret",
   "fixit", "diagnostics-parseable-fixits",
+  "syntax-only",
   "all",
   0
 };
@@ -239,6 +247,7 @@ static const int feature_types[] =
   feature_none,
   feature_caret, feature_caret,
   feature_fixit_parsable, feature_fixit_parsable,
+  feature_syntax_only,
   feature_all
 };
 
@@ -290,7 +299,7 @@ Operation modes:\n\
   -u, --update               apply fixes to the source grammar file and exit\n\
   -y, --yacc                 emulate POSIX Yacc\n\
   -W, --warnings[=CATEGORY]  report the warnings falling in CATEGORY\n\
-  -f, --feature[=FEATURE]    activate miscellaneous features\n\
+  -f, --feature[=FEATURES]   activate miscellaneous features\n\
 \n\
 "), stdout);
 
@@ -329,13 +338,13 @@ Output:\n\
 
       fputs (_("\
 Warning categories include:\n\
-  'midrule-values'    unset or unused midrule values\n\
-  'yacc'              incompatibilities with POSIX Yacc\n\
   'conflicts-sr'      S/R conflicts (enabled by default)\n\
   'conflicts-rr'      R/R conflicts (enabled by default)\n\
   'deprecated'        obsolete constructs\n\
   'empty-rule'        empty rules without %empty\n\
+  'midrule-values'    unset or unused midrule values\n\
   'precedence'        useless precedence and associativity\n\
+  'yacc'              incompatibilities with POSIX Yacc\n\
   'other'             all other warnings (enabled by default)\n\
   'all'               all the warnings except 'yacc'\n\
   'no-CATEGORY'       turn off warnings in CATEGORY\n\
@@ -356,10 +365,17 @@ THINGS is a list of comma separated words that can include:\n\
       putc ('\n', stdout);
 
       fputs (_("\
-FEATURE is a list of comma separated words that can include:\n\
-  'caret'        show errors with carets\n\
-  'all'          all of the above\n\
-  'none'         disable all of the above\n\
+FEATURES is a list of comma separated words that can include:\n\
+  'caret', 'diagnostics-show-caret'\n\
+    show errors with carets\n\
+  'fixit', 'diagnostics-parseable-fixits'\n\
+    show machine-readable fixes\n\
+  'syntax-only'\n\
+    do not generate any file\n\
+  'all'\n\
+    all of the above\n\
+  'none'\n\
+    disable all of the above\n\
   "), stdout);
 
       putc ('\n', stdout);
@@ -491,10 +507,12 @@ static char const short_options[] =
 /* Values for long options that do not have single-letter equivalents.  */
 enum
 {
-  LOCATIONS_OPTION = CHAR_MAX + 1,
-  PRINT_LOCALEDIR_OPTION,
+  COLOR_OPTION = CHAR_MAX + 1,
+  LOCATIONS_OPTION,
   PRINT_DATADIR_OPTION,
-  REPORT_FILE_OPTION
+  PRINT_LOCALEDIR_OPTION,
+  REPORT_FILE_OPTION,
+  STYLE_OPTION
 };
 
 static struct option const long_options[] =
@@ -521,7 +539,9 @@ static struct option const long_options[] =
   { "verbose",     no_argument,         0,   'v' },
 
   /* Hidden. */
-  { "trace",         optional_argument,   0,     'T' },
+  { "trace",       optional_argument,   0,  'T' },
+  { "color",       optional_argument,   0,  COLOR_OPTION },
+  { "style",       optional_argument,   0,  STYLE_OPTION },
 
   /* Output.  */
   { "defines",     optional_argument,   0,   'd' },
@@ -559,17 +579,46 @@ command_line_location (void)
 {
   location res;
   /* "<command line>" is used in GCC's messages about -D. */
-  boundary_set (&res.start, uniqstr_new ("<command line>"), optind - 1, -1);
+  boundary_set (&res.start, uniqstr_new ("<command line>"), optind - 1, -1, -1);
   res.end = res.start;
   return res;
 }
 
 
+/* Handle the command line options for color support.  Do it early, so
+   that error messages from getargs be also colored as per the user's
+   request.  This is consistent with the way GCC and Clang behave.  */
+
+static void
+getargs_colors (int argc, char *argv[])
+{
+  for (int i = 1; i < argc; i++)
+    {
+      const char *arg = argv[i];
+      if (STRPREFIX_LIT ("--color=", arg))
+        {
+          const char *color = arg + strlen ("--color=");
+          if (STREQ (color, "debug"))
+            color_debug = true;
+          else
+            handle_color_option (color);
+        }
+      else if (STRPREFIX_LIT ("--style=", arg))
+        {
+          const char *style = arg + strlen ("--style=");
+          handle_style_option (style);
+        }
+    }
+  complain_init_color ();
+}
+
+
 void
 getargs (int argc, char *argv[])
 {
-  int c;
+  getargs_colors (argc, argv);
 
+  int c;
   while ((c = getopt_long (argc, argv, short_options, long_options, NULL))
          != -1)
     switch (c)
@@ -646,8 +695,8 @@ getargs (int argc, char *argv[])
         defines_flag = true;
         if (optarg)
           {
-            free (spec_defines_file);
-            spec_defines_file = xstrdup (AS_FILE_NAME (optarg));
+            free (spec_header_file);
+            spec_header_file = xstrdup (AS_FILE_NAME (optarg));
           }
         break;
 
@@ -692,6 +741,7 @@ getargs (int argc, char *argv[])
 
       case 'u':
         update_flag = true;
+        feature_flag |= feature_syntax_only;
         break;
 
       case 'v':
@@ -712,6 +762,10 @@ getargs (int argc, char *argv[])
         yacc_loc = command_line_location ();
         break;
 
+      case COLOR_OPTION:
+        /* Handled in getargs_colors. */
+        break;
+
       case LOCATIONS_OPTION:
         muscle_percent_define_ensure ("locations",
                                       command_line_location (), true);
@@ -730,6 +784,10 @@ getargs (int argc, char *argv[])
         spec_verbose_file = xstrdup (AS_FILE_NAME (optarg));
         break;
 
+      case STYLE_OPTION:
+        /* Handled in getargs_colors. */
+        break;
+
       default:
         usage (EXIT_FAILURE);
       }

src/getargs.h

@@ -41,7 +41,7 @@ extern bool no_lines_flag;              /* for -l */
 extern bool token_table_flag;           /* for -k */
 extern location yacc_loc;               /* for -y */
 extern bool update_flag;                /* for -u */
-
+extern bool color_debug;                /* --color=debug. */
 /* GLR_PARSER is true if the input file says to use the GLR
    (Generalized LR) parser, and to output some additional information
    used by the GLR algorithm.  */
@@ -103,6 +103,8 @@ enum trace
     trace_m4        = 1 << 10, /**< M4 traces. */
     trace_muscles   = 1 << 11, /**< M4 definitions of the muscles. */
     trace_ielr      = 1 << 12, /**< IELR conversion. */
+    trace_closure   = 1 << 13, /**< Input/output of closure(). */
+    trace_locations = 1 << 14, /**< Full display of locations. */
     trace_all       = ~0       /**< All of the above.  */
   };
 /** What debug items bison displays during its run.  */
@@ -114,10 +116,11 @@ extern int trace_flag;
 
 enum feature
   {
-    feature_none           = 0,       /**< No additional feature.  */
-    feature_caret          = 1 << 0,  /**< Output errors with carets.  */
-    feature_fixit_parsable = 1 << 1,  /**< Issue instructions to fix the sources.  */
-    feature_all            = ~0       /**< All above features.  */
+    feature_none             = 0,      /**< No additional feature.  */
+    feature_caret            = 1 << 0, /**< Output errors with carets.  */
+    feature_fixit_parsable   = 1 << 1, /**< Issue instructions to fix the sources.  */
+    feature_syntax_only      = 1 << 2, /**< Don't generate output.  */
+    feature_all              = ~0      /**< All above features.  */
   };
 /** What additional features to use.  */
 extern int feature_flag;

src/gram.c

@@ -48,6 +48,34 @@ int max_user_token_number = 256;
 
 int required_version = 0;
 
+rule const *
+item_rule (item_number const *item)
+{
+  item_number const *sp = item;
+  while (0 <= *sp)
+    ++sp;
+  rule_number r = item_number_as_rule_number (*sp);
+  return &rules[r];
+}
+
+
+void
+item_print (item_number *item, rule const *previous_rule, FILE *out)
+{
+  rule const *r = item_rule (item);
+  rule_lhs_print (r, previous_rule ? previous_rule->lhs : NULL, out);
+
+  for (item_number *sp = r->rhs; sp < item; sp++)
+    fprintf (out, " %s", symbols[*sp]->tag);
+  fputs (" .", out);
+  if (0 <= *r->rhs)
+    for (item_number *sp = item; 0 <= *sp; ++sp)
+      fprintf (out, " %s", symbols[*sp]->tag);
+  else
+    fprintf (out, " %%empty");
+}
+
+
 bool
 rule_useful_in_grammar_p (rule const *r)
 {
@@ -66,6 +94,12 @@ rule_useless_in_parser_p (rule const *r)
   return !r->useful && rule_useful_in_grammar_p (r);
 }
 
+bool
+rule_useless_chain_p (rule const *r)
+{
+  return rule_rhs_length (r) == 1 && !r->action;
+}
+
 void
 rule_lhs_print (rule const *r, sym_content const *previous_lhs, FILE *out)
 {
@@ -86,7 +120,7 @@ size_t
 rule_rhs_length (rule const *r)
 {
   size_t res = 0;
-  for (item_number *rhsp = r->rhs; *rhsp >= 0; ++rhsp)
+  for (item_number *rhsp = r->rhs; 0 <= *rhsp; ++rhsp)
     ++res;
   return res;
 }
@@ -95,8 +129,8 @@ void
 rule_rhs_print (rule const *r, FILE *out)
 {
   if (0 <= *r->rhs)
-    for (item_number *rp = r->rhs; *rp >= 0; rp++)
-      fprintf (out, " %s", symbols[*rp]->tag);
+    for (item_number *rhsp = r->rhs; 0 <= *rhsp; ++rhsp)
+      fprintf (out, " %s", symbols[*rhsp]->tag);
   else
     fputs (" %empty", out);
 }
@@ -107,9 +141,9 @@ rule_rhs_print_xml (rule const *r, FILE *out, int level)
   if (*r->rhs >= 0)
     {
       xml_puts (out, level, "<rhs>");
-      for (item_number *rp = r->rhs; *rp >= 0; rp++)
+      for (item_number *rhsp = r->rhs; 0 <= *rhsp; ++rhsp)
         xml_printf (out, level + 1, "<symbol>%s</symbol>",
-                    xml_escape (symbols[*rp]->tag));
+                    xml_escape (symbols[*rhsp]->tag));
       xml_puts (out, level, "</rhs>");
     }
   else
@@ -120,6 +154,13 @@ rule_rhs_print_xml (rule const *r, FILE *out, int level)
     }
 }
 
+void
+rule_print (rule const *r, rule const *prev_rule, FILE *out)
+{
+  rule_lhs_print (r, prev_rule ? prev_rule->lhs : NULL, out);
+  rule_rhs_print (r, out);
+}
+
 void
 ritem_print (FILE *out)
 {
@@ -138,7 +179,7 @@ ritem_longest_rhs (void)
   int max = 0;
   for (rule_number r = 0; r < nrules; ++r)
     {
-      int length = rule_rhs_length (&rules[r]);
+      size_t length = rule_rhs_length (&rules[r]);
       if (length > max)
         max = length;
     }
@@ -151,7 +192,7 @@ grammar_rules_partial_print (FILE *out, const char *title,
                              rule_filter filter)
 {
   bool first = true;
-  sym_content *previous_lhs = NULL;
+  rule *previous_rule = NULL;
 
   /* rule # : LHS -> RHS */
   for (rule_number r = 0; r < nrules + nuseless_productions; r++)
@@ -160,13 +201,12 @@ grammar_rules_partial_print (FILE *out, const char *title,
         continue;
       if (first)
         fprintf (out, "%s\n\n", title);
-      else if (previous_lhs && previous_lhs != rules[r].lhs)
+      else if (previous_rule && previous_rule->lhs != rules[r].lhs)
         fputc ('\n', out);
       first = false;
-      rule_lhs_print (&rules[r], previous_lhs, out);
-      rule_rhs_print (&rules[r], out);
-      fprintf (out, "\n");
-      previous_lhs = rules[r].lhs;
+      rule_print (&rules[r], previous_rule, out);
+      fputc ('\n', out);
+      previous_rule = &rules[r];
     }
   if (!first)
     fputs ("\n\n", out);
@@ -189,13 +229,10 @@ grammar_rules_print_xml (FILE *out, int level)
         xml_puts (out, level + 1, "<rules>");
       first = false;
       {
-        char const *usefulness;
-        if (rule_useless_in_grammar_p (&rules[r]))
-          usefulness = "useless-in-grammar";
-        else if (rule_useless_in_parser_p (&rules[r]))
-          usefulness = "useless-in-parser";
-        else
-          usefulness = "useful";
+        char const *usefulness
+          = rule_useless_in_grammar_p (&rules[r]) ? "useless-in-grammar"
+          : rule_useless_in_parser_p (&rules[r])  ? "useless-in-parser"
+          :                                         "useful";
         xml_indent (out, level + 2);
         fprintf (out, "<rule number=\"%d\" usefulness=\"%s\"",
                  rules[r].number, usefulness);
@@ -238,39 +275,36 @@ grammar_dump (FILE *out, const char *title)
   fprintf (out, "Rules\n-----\n\n");
   {
     fprintf (out,
-             "Num (Prec, Assoc, Useful, Ritem Range) Lhs"
-             " -> Rhs (Ritem range) [Num]\n");
-    for (rule_number i = 0; i < nrules + nuseless_productions; i++)
+             "Num (Prec, Assoc, Useful, UselessChain) Lhs"
+             " -> (Ritem Range) Rhs\n");
+    for (rule_number i = 0; i < nrules + nuseless_productions; ++i)
       {
         rule const *rule_i = &rules[i];
         unsigned const rhs_itemno = rule_i->rhs - ritem;
-        /* Find the last RHS index in ritems. */
-        unsigned rhs_count = 0;
-        for (item_number *rp = rule_i->rhs; *rp >= 0; ++rp)
-          ++rhs_count;
-        fprintf (out, "%3d (%2d, %2d, %2d, %2u-%2u)   %2d ->",
+        unsigned length = rule_rhs_length (rule_i);
+        aver (item_number_as_rule_number (rule_i->rhs[length] == i));
+        fprintf (out, "%3d (%2d, %2d, %2s, %2s)   %2d -> (%2u-%2u)",
                  i,
                  rule_i->prec ? rule_i->prec->prec : 0,
                  rule_i->prec ? rule_i->prec->assoc : 0,
-                 rule_i->useful,
-                 rhs_itemno,
-                 rhs_itemno + rhs_count - 1,
-                 rule_i->lhs->number);
+                 rule_i->useful ? "t" : "f",
+                 rule_useless_chain_p (rule_i) ? "t" : "f",
+                 rule_i->lhs->number,
+                 rhs_itemno, rhs_itemno + length - 1);
         /* Dumped the RHS. */
-        for (item_number *rp = rule_i->rhs; *rp >= 0; ++rp)
-          fprintf (out, " %3d", *rp);
-        fprintf (out, "  [%d]\n",
-                 item_number_as_rule_number (rule_i->rhs[rhs_count+1]));
+        for (item_number *rhsp = rule_i->rhs; 0 <= *rhsp; ++rhsp)
+          fprintf (out, " %3d", *rhsp);
+        fputc ('\n', out);
       }
   }
   fprintf (out, "\n\n");
 
   fprintf (out, "Rules interpreted\n-----------------\n\n");
-  for (rule_number r = 0; r < nrules + nuseless_productions; r++)
+  for (rule_number r = 0; r < nrules + nuseless_productions; ++r)
     {
       fprintf (out, "%-5d  %s:", r, rules[r].lhs->symbol->tag);
       rule_rhs_print (&rules[r], out);
-      fprintf (out, "\n");
+      fputc ('\n', out);
     }
   fprintf (out, "\n\n");
 }
@@ -278,7 +312,7 @@ grammar_dump (FILE *out, const char *title)
 void
 grammar_rules_useless_report (const char *message)
 {
-  for (rule_number r = 0; r < nrules ; ++r)
+  for (rule_number r = 0; r < nrules; ++r)
     /* Don't complain about rules whose LHS is useless, we already
        complained about it.  */
     if (!reduce_nonterminal_useless_in_grammar (rules[r].lhs)

src/gram.h

@@ -75,7 +75,7 @@
 
    RULES[R].line -- the line where R was defined.
 
-   RULES[R].useful -- true iff the rule is used (i.e., false if thrown
+   RULES[R].useful -- whether the rule is used (i.e., false if thrown
    away by reduce).
 
    The right hand side is stored as symbol numbers in a portion of
@@ -146,7 +146,6 @@ item_number_is_symbol_number (item_number i)
 /* Rule numbers.  */
 typedef int rule_number;
 # define RULE_NUMBER_MAX INT_MAX
-extern rule_number nrules;
 
 static inline item_number
 rule_number_as_item_number (rule_number r)
@@ -166,6 +165,7 @@ item_number_is_rule_number (item_number i)
   return i < 0;
 }
 
+
 /*--------.
 | Rules.  |
 `--------*/
@@ -192,6 +192,7 @@ typedef struct
   /* This symbol was attached to the rule via %prec. */
   sym_content *precsym;
 
+  /* Location of the rhs.  */
   location location;
   bool useful;
   bool is_predicate;
@@ -202,26 +203,40 @@ typedef struct
   int expected_rr_conflicts;
 
   const char *action;
-  location action_location;
+  location action_loc;
 } rule;
 
+/* The used rules (size NRULES).  */
 extern rule *rules;
+extern rule_number nrules;
+
+/* Get the rule associated to this item.  ITEM points inside RITEM.  */
+rule const *item_rule (item_number const *item);
+
+/* Pretty-print this ITEM (as in the report).  ITEM points inside
+   RITEM.  PREVIOUS_RULE is used to see if the lhs is common, in which
+   case LHS is factored.  Passing NULL is fine.  */
+void item_print (item_number *item, rule const *previous_rule,
+                 FILE *out);
 
 /* A function that selects a rule.  */
 typedef bool (*rule_filter) (rule const *);
 
-/* Return true IFF the rule has a 'number' smaller than NRULES.  That is, it is
-   useful in the grammar.  */
+/* Whether the rule has a 'number' smaller than NRULES.  That is, it
+   is useful in the grammar.  */
 bool rule_useful_in_grammar_p (rule const *r);
 
-/* Return true IFF the rule has a 'number' higher than NRULES.  That is, it is
+/* Whether the rule has a 'number' higher than NRULES.  That is, it is
    useless in the grammar.  */
 bool rule_useless_in_grammar_p (rule const *r);
 
-/* Return true IFF the rule is not flagged as useful but is useful in the
+/* Whether the rule is not flagged as useful but is useful in the
    grammar.  In other words, it was discarded because of conflicts.  */
 bool rule_useless_in_parser_p (rule const *r);
 
+/* Whether the rule has a single RHS, and no user action. */
+bool rule_useless_chain_p (rule const *r);
+
 /* Print this rule's number and lhs on OUT.  If a PREVIOUS_LHS was
    already displayed (by a previous call for another rule), avoid
    useless repetitions.  */
@@ -229,12 +244,15 @@ void rule_lhs_print (rule const *r, sym_content const *previous_lhs,
                      FILE *out);
 void rule_lhs_print_xml (rule const *r, FILE *out, int level);
 
-/* Return the length of the RHS.  */
+/* The length of the RHS.  */
 size_t rule_rhs_length (rule const *r);
 
 /* Print this rule's RHS on OUT.  */
 void rule_rhs_print (rule const *r, FILE *out);
 
+/* Print this rule on OUT.  If a PREVIOUS_RULE was already displayed,
+   avoid useless repetitions of their LHS. */
+void rule_print (rule const *r, rule const *prev_rule, FILE *out);
 
 
 
@@ -252,7 +270,7 @@ extern int max_user_token_number;
 /* Dump RITEM for traces. */
 void ritem_print (FILE *out);
 
-/* Return the size of the longest rule RHS.  */
+/* The size of the longest rule RHS.  */
 size_t ritem_longest_rhs (void);
 
 /* Print the grammar's rules that match FILTER on OUT under TITLE.  */

src/graphviz.c

@@ -159,6 +159,9 @@ output_red (state const *s, reductions const *reds, FILE *fout)
   bitset no_reduce_set;
   no_reduce_bitset_init (s, &no_reduce_set);
 
+  rule *default_reduction
+    = yydefact[s->number] ? &rules[yydefact[s->number] - 1] : NULL;
+
   /* Two obstacks are needed: one for the enabled reductions, and one
      for the disabled reductions, because in the end we want two
      separate edges, even though in most cases only one will actually
@@ -171,11 +174,6 @@ output_red (state const *s, reductions const *reds, FILE *fout)
   const int source = s->number;
   for (int j = 0; j < reds->num; ++j)
     {
-      rule *default_reduction =
-        yydefact[s->number]
-        ? &rules[yydefact[s->number] - 1]
-        : NULL;
-
       bool defaulted = default_reduction && default_reduction == reds->rules[j];
 
       /* Build the lookahead tokens lists, one for enabled transitions

src/ielr.c

@@ -36,7 +36,36 @@
 #include "symtab.h"
 
 /** Records the value of the \%define variable lr.type.  */
-typedef enum { LR_TYPE__LALR, LR_TYPE__IELR, LR_TYPE__CANONICAL_LR } LrType;
+typedef enum
+  {
+   LR_TYPE__LR0,
+   LR_TYPE__LALR,
+   LR_TYPE__IELR,
+   LR_TYPE__CANONICAL_LR
+  } LrType;
+
+/* The user's requested LR type.  */
+static LrType
+lr_type_get (void)
+{
+  char *type = muscle_percent_define_get ("lr.type");
+  LrType res;
+  if (STREQ (type, "lr""(0)"))
+    res = LR_TYPE__LR0;
+  else if (STREQ (type, "lalr"))
+    res = LR_TYPE__LALR;
+  else if (STREQ (type, "ielr"))
+    res = LR_TYPE__IELR;
+  else if (STREQ (type, "canonical-lr"))
+    res = LR_TYPE__CANONICAL_LR;
+  else
+    {
+      aver (false);
+      abort ();
+    }
+  free (type);
+  return res;
+}
 
 /**
  * \post:
@@ -167,10 +196,7 @@ ielr_compute_internal_follow_edges (bitset ritem_sees_lookahead_set,
   relation_transpose (edgesp, ngotos);
 
   if (trace_flag & trace_ielr)
-    {
-      fprintf (stderr, "internal_follow_edges:\n");
-      relation_print (*edgesp, ngotos, stderr);
-    }
+    relation_print ("internal_follow_edges", *edgesp, ngotos, NULL, stderr);
 }
 
 /**
@@ -215,7 +241,7 @@ ielr_compute_follow_kernel_items (bitset ritem_sees_lookahead_set,
             && bitset_test (ritem_sees_lookahead_set, items[j]))
           bitset_set ((*follow_kernel_itemsp)[i], j);
     }
-  relation_digraph (internal_follow_edges, ngotos, follow_kernel_itemsp);
+  relation_digraph (internal_follow_edges, ngotos, *follow_kernel_itemsp);
 
   if (trace_flag & trace_ielr)
     {
@@ -271,12 +297,11 @@ ielr_compute_always_follows (goto_number ***edgesp,
       }
     free (edge_array);
   }
-  relation_digraph (*edgesp, ngotos, always_followsp);
+  relation_digraph (*edgesp, ngotos, *always_followsp);
 
   if (trace_flag & trace_ielr)
     {
-      fprintf (stderr, "always follow edges:\n");
-      relation_print (*edgesp, ngotos, stderr);
+      relation_print ("always follow edges", *edgesp, ngotos, NULL, stderr);
       fprintf (stderr, "always_follows:\n");
       debug_bitsetv (*always_followsp);
     }
@@ -540,7 +565,7 @@ typedef struct state_list
   /**
    * nextIsocore is the next state in a circularly linked-list of all states
    * with the same core.  The one originally computed by generate_states in
-   * LR0.c is lr0Isocore.
+   * lr0.c is lr0Isocore.
    */
   struct state_list *lr0Isocore;
   struct state_list *nextIsocore;
@@ -1051,41 +1076,40 @@ ielr_split_states (bitsetv follow_kernel_items, bitsetv always_follows,
     }
 }
 
+
 void
 ielr (void)
 {
-  LrType lr_type;
-
-  /* Examine user options.  */
-  {
-    char *type = muscle_percent_define_get ("lr.type");
-    if (STREQ (type, "lalr"))
-      lr_type = LR_TYPE__LALR;
-    else if (STREQ (type, "ielr"))
-      lr_type = LR_TYPE__IELR;
-    else if (STREQ (type, "canonical-lr"))
-      lr_type = LR_TYPE__CANONICAL_LR;
-    else
-      {
-        aver (false);
-        abort ();
-      }
-    free (type);
-  }
+  LrType lr_type = lr_type_get ();
 
   /* Phase 0: LALR(1).  */
-  timevar_push (tv_lalr);
-  if (lr_type == LR_TYPE__CANONICAL_LR)
-    set_goto_map ();
-  else
-    lalr ();
-  if (lr_type == LR_TYPE__LALR)
+  switch (lr_type)
     {
+    case LR_TYPE__LR0:
+      timevar_push (tv_lalr);
+      set_goto_map ();
+      timevar_pop (tv_lalr);
+      return;
+
+    case LR_TYPE__CANONICAL_LR:
+      timevar_push (tv_lalr);
+      set_goto_map ();
+      timevar_pop (tv_lalr);
+      break;
+
+    case LR_TYPE__LALR:
+      timevar_push (tv_lalr);
+      lalr ();
       bitsetv_free (goto_follows);
       timevar_pop (tv_lalr);
       return;
+
+    case LR_TYPE__IELR:
+      timevar_push (tv_lalr);
+      lalr ();
+      timevar_pop (tv_lalr);
+      break;
     }
-  timevar_pop (tv_lalr);
 
   {
     bitsetv follow_kernel_items;

src/lalr.c

@@ -28,18 +28,19 @@
 #include <bitset.h>
 #include <bitsetv.h>
 
-#include "LR0.h"
 #include "complain.h"
 #include "derives.h"
 #include "getargs.h"
 #include "gram.h"
 #include "lalr.h"
+#include "lr0.h"
 #include "muscle-tab.h"
 #include "nullable.h"
 #include "reader.h"
 #include "relation.h"
 #include "symtab.h"
 
+/* goto_map[nterm - NTOKENS] -> number of gotos.  */
 goto_number *goto_map = NULL;
 goto_number ngotos = 0;
 state_number *from_state = NULL;
@@ -53,8 +54,16 @@ typedef struct goto_list
   goto_number value;
 } goto_list;
 
+static goto_list *
+goto_list_new (goto_number value, struct goto_list *next)
+{
+  goto_list *res = xmalloc (sizeof *res);
+  res->next = next;
+  res->value = value;
+  return res;
+}
 
-/* LA is an NLA by NTOKENS matrix of bits.  LA[l, i] is 1 if the rule
+/* LA is an nLA by NTOKENS matrix of bits.  LA[l, i] is 1 if the rule
    LArule[l] is applicable in the appropriate state when the next
    token is symbol i.  If LA[l, i] and LA[l, j] are both 1 for i != j,
    it is a conflict.  */
@@ -63,31 +72,48 @@ static bitsetv LA = NULL;
 size_t nLA;
 
 
+/* "(p, A) includes (p', B)" iff
+   B → βAγ, γ nullable, and p'-- β --> p (i.e., state p' reaches p on label β).
+
+   Definition p.621 [DeRemer 1982].
+
+   INCLUDES[(p, A)] = [(p', B),...] */
 static goto_number **includes;
+
+/* "(q, A → ω) lookback (p, A)" iff state p reaches state q on label ω.
+
+   Definition p.621 [DeRemer 1982]. */
 static goto_list **lookback;
 
+static void
+goto_print (goto_number i, FILE *out)
+{
+  const state_number src = from_state[i];
+  const state_number dst = to_state[i];
+  symbol_number var = states[dst]->accessing_symbol;
+  fprintf (out,
+           "goto[%ld] = (%d, %s, %d)", i, src, symbols[var]->tag, dst);
+}
 
 void
 set_goto_map (void)
 {
-  goto_number *temp_map = xnmalloc (nvars + 1, sizeof *temp_map);
-
+  /* Count the number of gotos (ngotos) per nterm (goto_map). */
   goto_map = xcalloc (nvars + 1, sizeof *goto_map);
   ngotos = 0;
   for (state_number s = 0; s < nstates; ++s)
     {
-      transitions *sp = states[s]->transitions;
-      for (int i = sp->num - 1; i >= 0 && TRANSITION_IS_GOTO (sp, i); --i)
+      transitions *trans = states[s]->transitions;
+      for (int i = trans->num - 1; 0 <= i && TRANSITION_IS_GOTO (trans, i); --i)
         {
           ngotos++;
-
           /* Abort if (ngotos + 1) would overflow.  */
           aver (ngotos != GOTO_NUMBER_MAXIMUM);
-
-          goto_map[TRANSITION_SYMBOL (sp, i) - ntokens]++;
+          goto_map[TRANSITION_SYMBOL (trans, i) - ntokens]++;
         }
     }
 
+  goto_number *temp_map = xnmalloc (nvars + 1, sizeof *temp_map);
   {
     goto_number k = 0;
     for (symbol_number i = ntokens; i < nsyms; ++i)
@@ -108,21 +134,28 @@ set_goto_map (void)
 
   for (state_number s = 0; s < nstates; ++s)
     {
-      const transitions *sp = states[s]->transitions;
-      for (int i = sp->num - 1; i >= 0 && TRANSITION_IS_GOTO (sp, i); --i)
+      const transitions *trans = states[s]->transitions;
+      for (int i = trans->num - 1; 0 <= i && TRANSITION_IS_GOTO (trans, i); --i)
         {
-          goto_number k = temp_map[TRANSITION_SYMBOL (sp, i) - ntokens]++;
+          goto_number k = temp_map[TRANSITION_SYMBOL (trans, i) - ntokens]++;
           from_state[k] = s;
-          to_state[k] = sp->states[i]->number;
+          to_state[k] = trans->states[i]->number;
         }
     }
 
   free (temp_map);
+
+  if (trace_flag & trace_automaton)
+    for (int i = 0; i < ngotos; ++i)
+      {
+        goto_print (i, stderr);
+        fputc ('\n', stderr);
+      }
 }
 
 
 goto_number
-map_goto (state_number s0, symbol_number sym)
+map_goto (state_number src, symbol_number sym)
 {
   goto_number low = goto_map[sym - ntokens];
   goto_number high = goto_map[sym - ntokens + 1] - 1;
@@ -132,39 +165,62 @@ map_goto (state_number s0, symbol_number sym)
       aver (low <= high);
       goto_number middle = (low + high) / 2;
       state_number s = from_state[middle];
-      if (s == s0)
+      if (s == src)
         return middle;
-      else if (s < s0)
+      else if (s < src)
         low = middle + 1;
       else
         high = middle - 1;
     }
 }
 
-
+/* Print FOLLOWS for debugging.  */
 static void
-initialize_F (void)
+follows_print (const char* title, FILE *out)
+{
+  fprintf (out, "%s:\n", title);
+  for (goto_number i = 0; i < ngotos; ++i)
+    {
+      fputs ("    FOLLOWS[", out);
+      goto_print (i, out);
+      fputs ("] =", out);
+      bitset_iterator iter;
+      symbol_number sym;
+      BITSET_FOR_EACH (iter, goto_follows[i], sym, 0)
+        fprintf (out, " %s", symbols[sym]->tag);
+      fputc ('\n', out);
+    }
+  fputc ('\n', out);
+}
+
+/* Build goto_follows. */
+static void
+initialize_goto_follows (void)
 {
   goto_number **reads = xnmalloc (ngotos, sizeof *reads);
-  goto_number *edge = xnmalloc (ngotos + 1, sizeof *edge);
-  goto_number nedges = 0;
+  goto_number *edge = xnmalloc (ngotos, sizeof *edge);
 
   goto_follows = bitsetv_create (ngotos, ntokens, BITSET_FIXED);
 
   for (goto_number i = 0; i < ngotos; ++i)
     {
-      state_number stateno = to_state[i];
-      const transitions *sp = states[stateno]->transitions;
+      state_number dst = to_state[i];
+      const transitions *trans = states[dst]->transitions;
 
       int j;
-      FOR_EACH_SHIFT (sp, j)
-        bitset_set (goto_follows[i], TRANSITION_SYMBOL (sp, j));
+      FOR_EACH_SHIFT (trans, j)
+        bitset_set (goto_follows[i], TRANSITION_SYMBOL (trans, j));
 
-      for (; j < sp->num; ++j)
+      /* Gotos outgoing from DST. */
+      goto_number nedges = 0;
+      for (; j < trans->num; ++j)
         {
-          symbol_number sym = TRANSITION_SYMBOL (sp, j);
+          symbol_number sym = TRANSITION_SYMBOL (trans, j);
           if (nullable[sym - ntokens])
-            edge[nedges++] = map_goto (stateno, sym);
+            {
+              assert (nedges < ngotos);
+              edge[nedges++] = map_goto (dst, sym);
+            }
         }
 
       if (nedges == 0)
@@ -174,82 +230,164 @@ initialize_F (void)
           reads[i] = xnmalloc (nedges + 1, sizeof reads[i][0]);
           memcpy (reads[i], edge, nedges * sizeof edge[0]);
           reads[i][nedges] = END_NODE;
-          nedges = 0;
         }
     }
+  if (trace_flag & trace_automaton)
+    {
+      follows_print ("follows after shifts", stderr);
+      relation_print ("reads", reads, ngotos, goto_print, stderr);
+    }
 
-  relation_digraph (reads, ngotos, &goto_follows);
+  relation_digraph (reads, ngotos, goto_follows);
+  if (trace_flag & trace_automaton)
+    follows_print ("follows after read", stderr);
 
   for (goto_number i = 0; i < ngotos; ++i)
     free (reads[i]);
-
   free (reads);
   free (edge);
 }
 
 
+/* Find the state which LOOKBACK[LOOKBACK_INDEX] is about.  */
+static const state *
+lookback_find_state (int lookback_index)
+{
+  state *res = NULL;
+  for (int j = 0; j < nstates; ++j)
+    if (states[j]->reductions
+        && states[j]->reductions->lookahead_tokens)
+      {
+        if (states[j]->reductions->lookahead_tokens - LA > lookback_index)
+          /* Went too far. */
+          break;
+        else
+          res = states[j];
+      }
+  /* Pacify "potential null pointer dereference" warning.  */
+  if (!res)
+    abort ();
+  return res;
+}
+
+/* Print LOOKBACK for debugging.  */
 static void
-add_lookback_edge (state *s, rule *r, goto_number gotono)
+lookback_print (FILE *out)
+{
+  fputs ("lookback:\n", out);
+  for (int i = 0; i < nLA; ++i)
+    if (lookback[i])
+    {
+      fprintf (out, "   %3d = ", i);
+      const state *s = lookback_find_state (i);
+      int rnum = i - (s->reductions->lookahead_tokens - LA);
+      const rule *r = s->reductions->rules[rnum];
+      fprintf (out, "(%3d, ", s->number);
+      rule_print (r, NULL, out);
+      fputs (") ->", out);
+      for (goto_list *sp = lookback[i]; sp; sp = sp->next)
+        {
+          fputc (' ', out);
+          goto_print (sp->value, out);
+        }
+      fputc ('\n', out);
+    }
+  fputc ('\n', out);
+}
+
+/* Add (S, R) -> GOTONO to LOOKBACK.
+
+   "(q, A → ω) lookback (p, A)" iff state p reaches state q on label ω.
+
+   The goto number GOTONO, whose source is S (which is
+   inconsistent), */
+static void
+add_lookback_edge (state *s, rule const *r, goto_number gotono)
 {
   int ri = state_reduction_find (s, r);
-  goto_list *sp = xmalloc (sizeof *sp);
-  sp->next = lookback[(s->reductions->lookahead_tokens - LA) + ri];
-  sp->value = gotono;
-  lookback[(s->reductions->lookahead_tokens - LA) + ri] = sp;
+  int idx = (s->reductions->lookahead_tokens - LA) + ri;
+  lookback[idx] = goto_list_new (gotono, lookback[idx]);
 }
 
 
-
+/* Compute INCLUDES and LOOKBACK.  Corresponds to step E in Sec. 6 of
+   [DeRemer 1982].  */
 static void
 build_relations (void)
 {
-  goto_number *edge = xnmalloc (ngotos + 1, sizeof *edge);
-  state_number *states1 = xnmalloc (ritem_longest_rhs () + 1, sizeof *states1);
+  goto_number *edge = xnmalloc (ngotos, sizeof *edge);
+  state_number *path = xnmalloc (ritem_longest_rhs () + 1, sizeof *path);
 
   includes = xnmalloc (ngotos, sizeof *includes);
 
+  /* For each goto (from SRC to DST labeled by nterm VAR), iterate
+     over each rule with VAR as LHS, and find the path PATH from SRC
+     labeled with the RHS of the rule. */
   for (goto_number i = 0; i < ngotos; ++i)
     {
+      const state_number src = from_state[i];
+      const state_number dst = to_state[i];
+      symbol_number var = states[dst]->accessing_symbol;
+
+      /* Size of EDGE.  */
       int nedges = 0;
-      symbol_number symbol1 = states[to_state[i]]->accessing_symbol;
-
-      for (rule **rulep = derives[symbol1 - ntokens]; *rulep; rulep++)
+      for (rule **rulep = derives[var - ntokens]; *rulep; ++rulep)
         {
+          rule const *r = *rulep;
+          state *s = states[src];
+          path[0] = s->number;
+
+          /* Length of PATH.  */
           int length = 1;
-          item_number const *rp;
-          state *s = states[from_state[i]];
-          states1[0] = s->number;
-
-          for (rp = (*rulep)->rhs; ! item_number_is_rule_number (*rp); rp++)
+          for (item_number const *rp = r->rhs; 0 <= *rp; rp++)
             {
-              s = transitions_to (s->transitions,
-                                  item_number_as_symbol_number (*rp));
-              states1[length++] = s->number;
+              symbol_number sym = item_number_as_symbol_number (*rp);
+              s = transitions_to (s, sym);
+              path[length++] = s->number;
             }
 
+          /* S is the end of PATH.  */
           if (!s->consistent)
-            add_lookback_edge (s, *rulep, i);
+            add_lookback_edge (s, r, i);
 
-          length--;
-          bool done = false;
-          while (!done)
+          /* Walk back PATH from penultimate to beginning.
+
+             The "0 <= p" part is actually useless: each rhs ends in a
+             rule number (for which ISVAR(...) is false), and there is
+             a sentinel (ritem[-1]=0) before the first rhs.  */
+          for (int p = length - 2; 0 <= p && ISVAR (r->rhs[p]); --p)
             {
-              done = true;
-              /* Each rhs ends in a rule number, and there is a
-                 sentinel (ritem[-1]=0) before the first rhs, so it is safe to
-                 decrement RP here.  */
-              rp--;
-              if (ISVAR (*rp))
-                {
-                  /* Downcasting from item_number to symbol_number.  */
-                  edge[nedges++] = map_goto (states1[--length],
-                                             item_number_as_symbol_number (*rp));
-                  if (nullable[*rp - ntokens])
-                    done = false;
-                }
+              symbol_number sym = item_number_as_symbol_number (r->rhs[p]);
+              goto_number g = map_goto (path[p], sym);
+              /* Insert G if not already in EDGE.
+                 FIXME: linear search.  A bitset instead?  */
+              {
+                bool found = false;
+                for (int j = 0; !found && j < nedges; ++j)
+                  found = edge[j] == g;
+                if (!found)
+                  {
+                    assert (nedges < ngotos);
+                    edge[nedges++] = g;
+                  }
+              }
+              if (!nullable[sym - ntokens])
+                break;
             }
         }
 
+      if (trace_flag & trace_automaton)
+        {
+          goto_print (i, stderr);
+          fputs (" edges = ", stderr);
+          for (int j = 0; j < nedges; ++j)
+            {
+              fputc (' ', stderr);
+              goto_print (edge[j], stderr);
+            }
+          fputc ('\n', stderr);
+        }
+
       if (nedges == 0)
         includes[i] = NULL;
       else
@@ -262,21 +400,22 @@ build_relations (void)
     }
 
   free (edge);
-  free (states1);
+  free (path);
 
   relation_transpose (&includes, ngotos);
+  if (trace_flag & trace_automaton)
+    relation_print ("includes", includes, ngotos, goto_print, stderr);
 }
 
-
-
+/* Compute FOLLOWS from INCLUDES, and free INCLUDES.  */
 static void
-compute_FOLLOWS (void)
+compute_follows (void)
 {
-  relation_digraph (includes, ngotos, &goto_follows);
-
+  relation_digraph (includes, ngotos, goto_follows);
+  if (trace_flag & trace_sets)
+    follows_print ("follows after includes", stderr);
   for (goto_number i = 0; i < ngotos; ++i)
     free (includes[i]);
-
   free (includes);
 }
 
@@ -284,6 +423,9 @@ compute_FOLLOWS (void)
 static void
 compute_lookahead_tokens (void)
 {
+  if (trace_flag & trace_automaton)
+      lookback_print (stderr);
+
   for (size_t i = 0; i < nLA; ++i)
     for (goto_list *sp = lookback[i]; sp; sp = sp->next)
       bitset_or (LA[i], LA[i], goto_follows[sp->value]);
@@ -291,7 +433,6 @@ compute_lookahead_tokens (void)
   /* Free LOOKBACK. */
   for (size_t i = 0; i < nLA; ++i)
     LIST_FREE (goto_list, lookback[i]);
-
   free (lookback);
 }
 
@@ -304,16 +445,15 @@ compute_lookahead_tokens (void)
 static int
 state_lookahead_tokens_count (state *s, bool default_reduction_only_for_accept)
 {
-  int n_lookahead_tokens = 0;
-  reductions *rp = s->reductions;
-  transitions *sp = s->transitions;
+  const reductions *reds = s->reductions;
+  const transitions *trans = s->transitions;
 
   /* Transitions are only disabled during conflict resolution, and that
      hasn't happened yet, so there should be no need to check that
      transition 0 hasn't been disabled before checking if it is a shift.
      However, this check was performed at one time, so we leave it as an
      aver.  */
-  aver (sp->num == 0 || !TRANSITION_IS_DISABLED (sp, 0));
+  aver (trans->num == 0 || !TRANSITION_IS_DISABLED (trans, 0));
 
   /* We need a lookahead either to distinguish different reductions
      (i.e., there are two or more), or to distinguish a reduction from a
@@ -323,15 +463,13 @@ state_lookahead_tokens_count (state *s, bool default_reduction_only_for_accept)
      a lookahead token that makes sense there, and so no lookahead token
      should be read) if the user has otherwise disabled default
      reductions.  */
-  if (rp->num > 1
-      || (rp->num == 1 && sp->num && TRANSITION_IS_SHIFT (sp, 0))
-      || (rp->num == 1 && rp->rules[0]->number != 0
-          && default_reduction_only_for_accept))
-    n_lookahead_tokens += rp->num;
-  else
-    s->consistent = true;
+  s->consistent =
+    !(reds->num > 1
+      || (reds->num == 1 && trans->num && TRANSITION_IS_SHIFT (trans, 0))
+      || (reds->num == 1 && reds->rules[0]->number != 0
+          && default_reduction_only_for_accept));
 
-  return n_lookahead_tokens;
+  return s->consistent ? 0 : reds->num;
 }
 
 
@@ -385,32 +523,28 @@ initialize_LA (void)
 static void
 lookahead_tokens_print (FILE *out)
 {
-  fprintf (out, "Lookahead tokens: BEGIN\n");
+  fputs ("Lookaheads:\n", out);
   for (state_number i = 0; i < nstates; ++i)
     {
       const reductions *reds = states[i]->reductions;
-      int n_lookahead_tokens = 0;
-
-      if (reds->lookahead_tokens)
-        for (int j = 0; j < reds->num; ++j)
-          if (reds->lookahead_tokens[j])
-            ++n_lookahead_tokens;
-
-      fprintf (out, "State %d: %d lookahead tokens\n",
-               i, n_lookahead_tokens);
-
-      if (reds->lookahead_tokens)
-        for (int j = 0; j < reds->num; ++j)
-          {
-            bitset_iterator iter;
-            int k;
-            BITSET_FOR_EACH (iter, reds->lookahead_tokens[j], k, 0)
-              fprintf (out, "   on %d (%s) -> rule %d\n",
-                       k, symbols[k]->tag,
-                       reds->rules[j]->number);
-          }
+      if (reds->num)
+        {
+          fprintf (out, "  State %d:\n", i);
+          for (int j = 0; j < reds->num; ++j)
+            {
+              fprintf (out, "    rule %d:", reds->rules[j]->number);
+              if (reds->lookahead_tokens)
+              {
+                bitset_iterator iter;
+                int k;
+                BITSET_FOR_EACH (iter, reds->lookahead_tokens[j], k, 0)
+                  fprintf (out, " %s", symbols[k]->tag);
+              }
+              fputc ('\n', out);
+            }
+        }
     }
-  fprintf (out, "Lookahead tokens: END\n");
+  fputc ('\n', out);
 }
 
 void
@@ -418,10 +552,10 @@ lalr (void)
 {
   initialize_LA ();
   set_goto_map ();
-  initialize_F ();
+  initialize_goto_follows ();
   lookback = xcalloc (nLA, sizeof *lookback);
   build_relations ();
-  compute_FOLLOWS ();
+  compute_follows ();
   compute_lookahead_tokens ();
 
   if (trace_flag & trace_sets)

src/lalr.h

@@ -83,8 +83,8 @@ typedef size_t goto_number;
 /** Index into #from_state and #to_state.
 
    All the transitions that accept a particular variable are grouped
-   together and GOTO_MAP[I - NTOKENS] is the index in FROM_STATE and
-   TO_STATE of the first of them.  */
+   together in FROM_STATE and TO_STATE, with indexes from GOTO_MAP[I -
+   NTOKENS] to GOTO_MAP[I - NTOKENS + 1] - 1 (including both).  */
 extern goto_number *goto_map;
 
 /** The size of #from_state and #to_state.  */
@@ -96,11 +96,10 @@ extern state_number *from_state;
 /** State number it leads to.  */
 extern state_number *to_state;
 
-/** Map a state/symbol pair into its numeric representation.  */
-goto_number map_goto (state_number s0, symbol_number sym);
+/** The number of the goto from state SRC labeled with nterm SYM.  */
+goto_number map_goto (state_number src, symbol_number sym);
 
 /* goto_follows[i] is the set of tokens following goto i.  */
 extern bitsetv goto_follows;
 
-
 #endif /* !LALR_H_ */

src/local.mk

@@ -30,8 +30,6 @@ src_bison_SOURCES =                             \
   src/AnnotationList.h                          \
   src/InadequacyList.c                          \
   src/InadequacyList.h                          \
-  src/LR0.c                                     \
-  src/LR0.h                                     \
   src/Sbitset.c                                 \
   src/Sbitset.h                                 \
   src/assoc.c                                   \
@@ -55,12 +53,14 @@ src_bison_SOURCES =                             \
   src/gram.h                                    \
   src/graphviz.c                                \
   src/graphviz.h                                \
-  src/lalr.c                                    \
-  src/lalr.h                                    \
   src/ielr.c                                    \
   src/ielr.h                                    \
+  src/lalr.c                                    \
+  src/lalr.h                                    \
   src/location.c                                \
   src/location.h                                \
+  src/lr0.c                                     \
+  src/lr0.h                                     \
   src/main.c                                    \
   src/muscle-tab.c                              \
   src/muscle-tab.h                              \
@@ -71,12 +71,12 @@ src_bison_SOURCES =                             \
   src/output.c                                  \
   src/output.h                                  \
   src/parse-gram.y                              \
+  src/print-graph.c                             \
+  src/print-graph.h                             \
   src/print-xml.c                               \
   src/print-xml.h                               \
   src/print.c                                   \
   src/print.h                                   \
-  src/print_graph.c                             \
-  src/print_graph.h                             \
   src/reader.c                                  \
   src/reader.h                                  \
   src/reduce.c                                  \
@@ -126,6 +126,7 @@ src_bison_LDADD =                               \
   $(LIBTHREAD)                                  \
   $(LIB_CLOCK_GETTIME)                          \
   $(LIB_GETHRXTIME)                             \
+  $(LIBTEXTSTYLE)                               \
   lib/libbison.a
 
 

src/location.c

@@ -23,33 +23,27 @@
 
 #include <mbswidth.h>
 #include <quotearg.h>
+#include <stdio.h>    /* fileno */
+#include <sys/stat.h> /* fstat */
 
 #include "complain.h"
+#include "getargs.h"
 #include "location.h"
 
-location const empty_location = EMPTY_LOCATION_INIT;
+location const empty_loc = EMPTY_LOCATION_INIT;
 
 /* If BUF is null, add BUFSIZE (which in this case must be less than
    INT_MAX) to COLUMN; otherwise, add mbsnwidth (BUF, BUFSIZE, 0) to
-   COLUMN.  If an overflow occurs, or might occur but is undetectable,
-   return INT_MAX.  Assume COLUMN is nonnegative.  */
+   COLUMN.  If an overflow occurs, return INT_MAX.  */
 
 static inline int
 add_column_width (int column, char const *buf, size_t bufsize)
 {
-  size_t width;
-  unsigned remaining_columns = INT_MAX - column;
-
-  if (buf)
-    {
-      if (INT_MAX / 2 <= bufsize)
-        return INT_MAX;
-      width = mbsnwidth (buf, bufsize, 0);
-    }
-  else
-    width = bufsize;
-
-  return width <= remaining_columns ? column + width : INT_MAX;
+  int width
+    = buf ? mbsnwidth (buf, bufsize, 0)
+    : INT_MAX <= bufsize ? INT_MAX
+    : bufsize;
+  return column <= INT_MAX - width ? column + width : INT_MAX;
 }
 
 /* Set *LOC and adjust scanner cursor to account for token TOKEN of
@@ -60,18 +54,20 @@ location_compute (location *loc, boundary *cur, char const *token, size_t size)
 {
   int line = cur->line;
   int column = cur->column;
+  int byte = cur->byte;
   char const *p0 = token;
   char const *p = token;
   char const *lim = token + size;
 
   loc->start = *cur;
 
-  for (p = token; p < lim; p++)
+  for (p = token; p < lim; ++p)
     switch (*p)
       {
       case '\n':
         line += line < INT_MAX;
         column = 1;
+        byte = 1;
         p0 = p + 1;
         break;
 
@@ -79,14 +75,17 @@ location_compute (location *loc, boundary *cur, char const *token, size_t size)
         column = add_column_width (column, p0, p - p0);
         column = add_column_width (column, NULL, 8 - ((column - 1) & 7));
         p0 = p + 1;
+        byte += byte < INT_MAX;
         break;
 
       default:
+        byte += byte < INT_MAX;
         break;
       }
 
   cur->line = line;
   cur->column = column = add_column_width (column, p0, p - p0);
+  cur->byte = byte;
 
   loc->end = *cur;
 
@@ -94,44 +93,62 @@ location_compute (location *loc, boundary *cur, char const *token, size_t size)
     complain (loc, Wother, _("line number overflow"));
   if (column == INT_MAX && loc->start.column != INT_MAX)
     complain (loc, Wother, _("column number overflow"));
+  if (byte == INT_MAX && loc->start.byte != INT_MAX)
+    complain (loc, Wother, _("byte number overflow"));
 }
 
+static unsigned
+boundary_print (boundary const *b, FILE *out)
+{
+  return fprintf (out, "%s:%d.%d@%d",
+                  quotearg_n_style (3, escape_quoting_style, b->file),
+                  b->line, b->column, b->byte);
+}
 
 unsigned
 location_print (location loc, FILE *out)
 {
   unsigned res = 0;
-  int end_col = 0 != loc.end.column ? loc.end.column - 1 : 0;
-  res += fprintf (out, "%s",
-                  quotearg_n_style (3, escape_quoting_style, loc.start.file));
-  if (0 <= loc.start.line)
+  if (trace_flag & trace_locations)
     {
-      res += fprintf (out, ":%d", loc.start.line);
-      if (0 <= loc.start.column)
-        res += fprintf (out, ".%d", loc.start.column);
+      res += boundary_print (&loc.start, out);
+      res += fprintf (out, "-");
+      res += boundary_print (&loc.end, out);
     }
-  if (loc.start.file != loc.end.file)
+  else
     {
-      res += fprintf (out, "-%s",
-                      quotearg_n_style (3, escape_quoting_style,
-                                        loc.end.file));
-      if (0 <= loc.end.line)
+      int end_col = 0 != loc.end.column ? loc.end.column - 1 : 0;
+      res += fprintf (out, "%s",
+                      quotearg_n_style (3, escape_quoting_style, loc.start.file));
+      if (0 <= loc.start.line)
         {
-          res += fprintf (out, ":%d", loc.end.line);
-          if (0 <= end_col)
-            res += fprintf (out, ".%d", end_col);
+          res += fprintf (out, ":%d", loc.start.line);
+          if (0 <= loc.start.column)
+            res += fprintf (out, ".%d", loc.start.column);
         }
-    }
-  else if (0 <= loc.end.line)
-    {
-      if (loc.start.line < loc.end.line)
+      if (loc.start.file != loc.end.file)
         {
-          res += fprintf (out, "-%d", loc.end.line);
-          if (0 <= end_col)
-            res += fprintf (out, ".%d", end_col);
+          res += fprintf (out, "-%s",
+                          quotearg_n_style (3, escape_quoting_style,
+                                            loc.end.file));
+          if (0 <= loc.end.line)
+            {
+              res += fprintf (out, ":%d", loc.end.line);
+              if (0 <= end_col)
+                res += fprintf (out, ".%d", end_col);
+            }
+        }
+      else if (0 <= loc.end.line)
+        {
+          if (loc.start.line < loc.end.line)
+            {
+              res += fprintf (out, "-%d", loc.end.line);
+              if (0 <= end_col)
+                res += fprintf (out, ".%d", end_col);
+            }
+          else if (0 <= end_col && loc.start.column < end_col)
+            res += fprintf (out, "-%d", end_col);
         }
-      else if (0 <= end_col && loc.start.column < end_col)
-        res += fprintf (out, "-%d", end_col);
     }
 
   return res;
@@ -140,34 +157,63 @@ location_print (location loc, FILE *out)
 
 /* Persistent data used by location_caret to avoid reopening and rereading the
    same file all over for each error.  */
-struct caret_info
+static struct
 {
   FILE *source;
+  /* The last file we tried to open.  If non NULL, but SOURCE is NULL,
+     it means this file is special and should not be quoted. */
+  uniqstr file;
   size_t line;
+  /* Offset in SOURCE where line LINE starts.  */
   size_t offset;
-};
-
-static struct caret_info caret_info = { NULL, 1, 0 };
+} caret_info;
 
 void
-cleanup_caret ()
+caret_free ()
 {
   if (caret_info.source)
-    fclose (caret_info.source);
-  caret_info.source = NULL;
-  caret_info.line = 1;
-  caret_info.offset = 0;
+    {
+      fclose (caret_info.source);
+      caret_info.source = NULL;
+    }
 }
 
 void
-location_caret (location loc, FILE *out)
+location_caret (location loc, const char *style, FILE *out)
 {
-  /* FIXME: find a way to support multifile locations, and only open once each
-     file. That would make the procedure future-proof.  */
-  if (! (caret_info.source
-         || (caret_info.source = fopen (loc.start.file, "r")))
-      || loc.start.column == -1 || loc.start.line == -1)
+  if (loc.start.column == -1 || loc.start.line == -1)
     return;
+  /* If a different source than before, close and let the rest open
+     the new one. */
+  if (caret_info.file && caret_info.file != loc.start.file)
+    {
+      caret_free ();
+      caret_info.file = NULL;
+    }
+  if (!caret_info.file)
+    {
+      caret_info.file = loc.start.file;
+      if ((caret_info.source = fopen (caret_info.file, "r")))
+        {
+          /* If the file is not regular (imagine #line 1 "/dev/stdin"
+             in the input file for instance), don't try to quote the
+             source.  Keep caret_info.file set so that we don't try to
+             open it again, but leave caret_info.source NULL so that
+             we don't try to quote it. */
+          struct stat buf;
+          if (fstat (fileno (caret_info.source), &buf) == 0
+              && buf.st_mode & S_IFREG)
+            {
+              caret_info.line = 1;
+              caret_info.offset = 0;
+            }
+          else
+            caret_free ();
+        }
+    }
+  if (!caret_info.source)
+    return;
+
 
   /* If the line we want to quote is seekable (the same line as the previous
      location), just seek it. If it was a previous line, we lost track of it,
@@ -192,25 +238,44 @@ location_caret (location loc, FILE *out)
     int c = getc (caret_info.source);
     if (c != EOF)
       {
-        /* Quote the file, indent by a single column.  */
-        putc (' ', out);
-        do
-          putc (c, out);
-        while ((c = getc (caret_info.source)) != EOF && c != '\n');
+        /* Quote the file (at most the first line in the case of
+           multiline locations).  */
+        fprintf (out, "%5d | ", loc.start.line);
+        bool single_line = loc.start.line == loc.end.line;
+        /* Consider that single point location (with equal boundaries)
+           actually denote the character that they follow.  */
+        int byte_end = loc.end.byte +
+          (single_line && loc.start.byte == loc.end.byte);
+        /* Byte number.  */
+        int byte = 1;
+        while (c != EOF && c != '\n')
+          {
+            if (byte == loc.start.byte)
+              begin_use_class (style, out);
+            fputc (c, out);
+            c = getc (caret_info.source);
+            ++byte;
+            if (single_line
+                ? byte == byte_end
+                : c == '\n' || c == EOF)
+              end_use_class (style, out);
+          }
         putc ('\n', out);
 
         {
-          /* The caret of a multiline location ends with the first line.  */
-          size_t len = loc.start.line != loc.end.line
-            ? ftell (caret_info.source) - caret_info.offset
-            : loc.end.column;
-          int i;
-
-          /* Print the carets (at least one), with the same indent as above.*/
-          fprintf (out, " %*s", loc.start.column - 1, "");
-          for (i = loc.start.column; i == loc.start.column || i < len; ++i)
-            putc (i == loc.start.column ? '^' : '~', out);
-          }
+          /* Print the carets with the same indentation as above.  */
+          fprintf (out, "      | %*s", loc.start.column - 1, "");
+          begin_use_class (style, out);
+          putc ('^', out);
+          /* Underlining a multiline location ends with the first
+             line.  */
+          int len = single_line
+            ? loc.end.column
+            : ftell (caret_info.source) - caret_info.offset;
+          for (int i = loc.start.column + 1; i < len; ++i)
+            putc ('~', out);
+          end_use_class (style, out);
+        }
         putc ('\n', out);
       }
   }
@@ -224,17 +289,29 @@ location_empty (location loc)
 }
 
 void
-boundary_set_from_string (boundary *bound, char *loc_str)
+boundary_set_from_string (boundary *bound, char *str)
 {
-  /* Must search in reverse since the file name field may
-   * contain '.' or ':'.  */
-  char *delim = strrchr (loc_str, '.');
-  aver (delim);
-  *delim = '\0';
-  bound->column = atoi (delim+1);
-  delim = strrchr (loc_str, ':');
-  aver (delim);
-  *delim = '\0';
-  bound->line = atoi (delim+1);
-  bound->file = uniqstr_new (loc_str);
+  /* Must search in reverse since the file name field may contain '.'
+     or ':'.  */
+  char *at = strrchr (str, '@');
+  if (at)
+    {
+      *at = '\0';
+      bound->byte = atoi (at+1);
+    }
+  {
+    char *dot = strrchr (str, '.');
+    aver (dot);
+    *dot = '\0';
+    bound->column = atoi (dot+1);
+    if (!at)
+      bound->byte = bound->column;
+  }
+  {
+    char *colon = strrchr (str, ':');
+    aver (colon);
+    *colon = '\0';
+    bound->line = atoi (colon+1);
+  }
+  bound->file = uniqstr_new (str);
 }

src/location.h

@@ -49,15 +49,21 @@ typedef struct
   */
   int column;
 
+  /* If nonnegative, (origin-0) bytes number in the current line.
+     Never displayed, used when printing error messages with colors to
+     know where colors start and ends.  */
+  int byte;
+
 } boundary;
 
-/* Set the position of \a a. */
+/* Set the position of \a p. */
 static inline void
-boundary_set (boundary *b, const char *f, int l, int c)
+boundary_set (boundary *p, const char *f, int l, int c, int b)
 {
-  b->file = f;
-  b->line = l;
-  b->column = c;
+  p->file = f;
+  p->line = l;
+  p->column = c;
+  p->byte = b;
 }
 
 /* Return -1, 0, 1, depending whether a is before, equal, or
@@ -95,8 +101,8 @@ typedef struct
 
 # define GRAM_LTYPE location
 
-# define EMPTY_LOCATION_INIT {{NULL, 0, 0}, {NULL, 0, 0}}
-extern location const empty_location;
+# define EMPTY_LOCATION_INIT {{NULL, 0, 0, 0}, {NULL, 0, 0, 0}}
+extern location const empty_loc;
 
 /* Set *LOC and adjust scanner cursor to account for token TOKEN of
    size SIZE.  */
@@ -110,10 +116,10 @@ unsigned location_print (location loc, FILE *out);
 
 /* Free any allocated resources and close any open file handles that are
    left-over by the usage of location_caret.  */
-void cleanup_caret (void);
+void caret_free (void);
 
 /* Output to OUT the line and caret corresponding to location LOC.  */
-void location_caret (location loc, FILE *out);
+void location_caret (location loc, const char* style, FILE *out);
 
 /* Return -1, 0, 1, depending whether a is before, equal, or
    after b.  */
@@ -129,7 +135,8 @@ location_cmp (location a, location b)
 /* Whether this is the empty location.  */
 bool location_empty (location loc);
 
-/* LOC_STR must be formatted as 'file:line.column', it will be modified.  */
-void boundary_set_from_string (boundary *bound, char *loc_str);
+/* STR must be formatted as 'file:line.column@byte' or 'file:line.column',
+   it will be modified.  */
+void boundary_set_from_string (boundary *bound, char *str);
 
 #endif /* ! defined LOCATION_H_ */

src/lr0.c

@@ -27,12 +27,12 @@
 
 #include <bitset.h>
 
-#include "LR0.h"
 #include "closure.h"
 #include "complain.h"
 #include "getargs.h"
 #include "gram.h"
 #include "lalr.h"
+#include "lr0.h"
 #include "reader.h"
 #include "reduce.h"
 #include "state.h"
@@ -47,6 +47,16 @@ typedef struct state_list
 static state_list *first_state = NULL;
 static state_list *last_state = NULL;
 
+/* Print CORE for debugging. */
+static void
+core_print (size_t core_size, item_number *core, FILE *out)
+{
+  for (int i = 0; i < core_size; ++i)
+    {
+      item_print (ritem + core[i], NULL, out);
+      fputc ('\n', out);
+    }
+}
 
 /*------------------------------------------------------------------.
 | A state was just discovered from another state.  Queue it for     |
@@ -57,14 +67,14 @@ static state *
 state_list_append (symbol_number sym, size_t core_size, item_number *core)
 {
   state_list *node = xmalloc (sizeof *node);
-  state *s = state_new (sym, core_size, core);
+  state *res = state_new (sym, core_size, core);
 
   if (trace_flag & trace_automaton)
     fprintf (stderr, "state_list_append (state = %d, symbol = %d (%s))\n",
              nstates, sym, symbols[sym]->tag);
 
   node->next = NULL;
-  node->state = s;
+  node->state = res;
 
   if (!first_state)
     first_state = node;
@@ -72,17 +82,28 @@ state_list_append (symbol_number sym, size_t core_size, item_number *core)
     last_state->next = node;
   last_state = node;
 
-  return s;
+  return res;
 }
 
-static int nshifts;
-static symbol_number *shift_symbol;
+/* Symbols that can be "shifted" (including non terminals) from the
+   current state.  */
+bitset shift_symbol;
 
 static rule **redset;
+/* For the current state, the list of pointers to states that can be
+   reached via a shift/goto.  Could be indexed by the reaching symbol,
+   but labels of incoming transitions can be recovered by the state
+   itself.  */
 static state **shiftset;
 
+
+/* KERNEL_BASE[symbol-number] -> list of item numbers (offsets inside
+   RITEM) of lenngth KERNEL_SIZE[symbol-number]. */
 static item_number **kernel_base;
 static int *kernel_size;
+
+/* A single dimension array that serves as storage for
+   KERNEL_BASE.  */
 static item_number *kernel_items;
 
 
@@ -98,10 +119,11 @@ allocate_itemsets (void)
                                   sizeof *symbol_count);
 
   for (rule_number r = 0; r < nrules; ++r)
-    for (item_number *rhsp = rules[r].rhs; *rhsp >= 0; ++rhsp)
+    for (item_number *rhsp = rules[r].rhs; 0 <= *rhsp; ++rhsp)
       {
-        count++;
-        symbol_count[*rhsp]++;
+        symbol_number sym = item_number_as_symbol_number (*rhsp);
+        count += 1;
+        symbol_count[sym] += 1;
       }
 
   /* See comments before new_itemsets.  All the vectors of items
@@ -124,6 +146,17 @@ allocate_itemsets (void)
   kernel_size = xnmalloc (nsyms, sizeof *kernel_size);
 }
 
+/* Print the current kernel (in KERNEL_BASE). */
+static void
+kernel_print (FILE *out)
+{
+  for (symbol_number i = 0; i < nsyms; ++i)
+    if (kernel_size[i])
+      {
+        fprintf (out, "kernel[%s] =\n", symbols[i]->tag);
+        core_print (kernel_size[i], kernel_base[i], out);
+      }
+}
 
 static void
 allocate_storage (void)
@@ -133,14 +166,14 @@ allocate_storage (void)
   shiftset = xnmalloc (nsyms, sizeof *shiftset);
   redset = xnmalloc (nrules, sizeof *redset);
   state_hash_new ();
-  shift_symbol = xnmalloc (nsyms, sizeof *shift_symbol);
+  shift_symbol = bitset_create (nsyms, BITSET_FIXED);
 }
 
 
 static void
 free_storage (void)
 {
-  free (shift_symbol);
+  bitset_free (shift_symbol);
   free (redset);
   free (shiftset);
   free (kernel_base);
@@ -152,44 +185,44 @@ free_storage (void)
 
 
 
-/*---------------------------------------------------------------.
-| Find which symbols can be shifted in S, and for each one       |
-| record which items would be active after that shift.  Uses the |
-| contents of itemset.                                           |
-|                                                                |
-| shift_symbol is set to a vector of the symbols that can be     |
-| shifted.  For each symbol in the grammar, kernel_base[symbol]  |
-| points to a vector of item numbers activated if that symbol is |
-| shifted, and kernel_size[symbol] is their numbers.             |
-|                                                                |
-| itemset is sorted on item index in ritem, which is sorted on   |
-| rule number.  Compute each kernel_base[symbol] with the same   |
-| sort.                                                          |
-`---------------------------------------------------------------*/
+/*------------------------------------------------------------------.
+| Find which term/nterm symbols can be "shifted" in S, and for each |
+| one record which items would be active after that transition.     |
+| Uses the contents of itemset.                                     |
+|                                                                   |
+| shift_symbol is a bitset of the term/nterm symbols that can be    |
+| shifted.  For each symbol in the grammar, kernel_base[symbol]     |
+| points to a vector of item numbers activated if that symbol is    |
+| shifted, and kernel_size[symbol] is their numbers.                |
+|                                                                   |
+| itemset is sorted on item index in ritem, which is sorted on rule |
+| number.  Compute each kernel_base[symbol] with the same sort.     |
+`------------------------------------------------------------------*/
 
 static void
 new_itemsets (state *s)
 {
   if (trace_flag & trace_automaton)
-    fprintf (stderr, "Entering new_itemsets, state = %d\n", s->number);
+    fprintf (stderr, "new_itemsets: begin: state = %d\n", s->number);
 
   memset (kernel_size, 0, nsyms * sizeof *kernel_size);
 
-  nshifts = 0;
+  bitset_zero (shift_symbol);
 
   for (size_t i = 0; i < nitemset; ++i)
     if (item_number_is_symbol_number (ritem[itemset[i]]))
       {
         symbol_number sym = item_number_as_symbol_number (ritem[itemset[i]]);
-        if (!kernel_size[sym])
-          {
-            shift_symbol[nshifts] = sym;
-            nshifts++;
-          }
-
+        bitset_set (shift_symbol, sym);
         kernel_base[sym][kernel_size[sym]] = itemset[i] + 1;
         kernel_size[sym]++;
       }
+
+  if (trace_flag & trace_automaton)
+    {
+      kernel_print (stderr);
+      fprintf (stderr, "new_itemsets: end: state = %d\n\n", s->number);
+    }
 }
 
 
@@ -204,8 +237,12 @@ static state *
 get_state (symbol_number sym, size_t core_size, item_number *core)
 {
   if (trace_flag & trace_automaton)
-    fprintf (stderr, "Entering get_state, symbol = %d (%s)\n",
-             sym, symbols[sym]->tag);
+    {
+      fprintf (stderr, "Entering get_state, symbol = %d (%s), core:\n",
+               sym, symbols[sym]->tag);
+      core_print (core_size, core, stderr);
+      fputc ('\n', stderr);
+    }
 
   state *s = state_hash_lookup (core_size, core);
   if (!s)
@@ -228,24 +265,19 @@ static void
 append_states (state *s)
 {
   if (trace_flag & trace_automaton)
-    fprintf (stderr, "Entering append_states, state = %d\n", s->number);
+    fprintf (stderr, "append_states: begin: state = %d\n", s->number);
 
-  /* First sort shift_symbol into increasing order.  */
-
-  for (int i = 1; i < nshifts; i++)
+  bitset_iterator iter;
+  symbol_number sym;
+  int i = 0;
+  BITSET_FOR_EACH (iter, shift_symbol, sym, 0)
     {
-      const symbol_number sym = shift_symbol[i];
-      int j = i;
-      for (; 0 < j && sym < shift_symbol[j - 1]; j--)
-        shift_symbol[j] = shift_symbol[j - 1];
-      shift_symbol[j] = sym;
-    }
-
-  for (int i = 0; i < nshifts; i++)
-    {
-      const symbol_number sym = shift_symbol[i];
       shiftset[i] = get_state (sym, kernel_size[sym], kernel_base[sym]);
+      ++i;
     }
+
+  if (trace_flag & trace_automaton)
+    fprintf (stderr, "append_states: end: state = %d\n", s->number);
 }
 
 
@@ -324,7 +356,7 @@ void
 generate_states (void)
 {
   allocate_storage ();
-  new_closure (nritems);
+  closure_new (nritems);
 
   /* Create the initial state.  The 0 at the lhs is the index of the
      item of this initial rule.  */
@@ -344,18 +376,17 @@ generate_states (void)
       closure (s->items, s->nitems);
       /* Record the reductions allowed out of this state.  */
       save_reductions (s);
-      /* Find the itemsets of the states that shifts can reach.  */
+      /* Find the itemsets of the states that shifts/gotos can reach.  */
       new_itemsets (s);
       /* Find or create the core structures for those states.  */
       append_states (s);
 
       /* Create the shifts structures for the shifts to those states,
          now that the state numbers transitioning to are known.  */
-      state_transitions_set (s, nshifts, shiftset);
+      state_transitions_set (s, bitset_count (shift_symbol), shiftset);
     }
 
   /* discard various storage */
-  free_closure ();
   free_storage ();
 
   /* Set up STATES. */

src/main.c

@@ -23,14 +23,14 @@
 
 #include <bitset.h>
 #include <bitset/stats.h>
+#include <closeout.h>
 #include <configmake.h>
 #include <progname.h>
+#include <quote.h>
 #include <quotearg.h>
 #include <relocatable.h> /* relocate2 */
 #include <timevar.h>
 
-#include "LR0.h"
-#include "closeout.h"
 #include "complain.h"
 #include "conflicts.h"
 #include "derives.h"
@@ -38,15 +38,15 @@
 #include "fixits.h"
 #include "getargs.h"
 #include "gram.h"
-#include "lalr.h"
 #include "ielr.h"
+#include "lalr.h"
+#include "lr0.h"
 #include "muscle-tab.h"
 #include "nullable.h"
 #include "output.h"
-#include "print.h"
-#include "print_graph.h"
+#include "print-graph.h"
 #include "print-xml.h"
-#include <quote.h>
+#include "print.h"
 #include "reader.h"
 #include "reduce.h"
 #include "scan-code.h"
@@ -66,11 +66,12 @@ main (int argc, char *argv[])
   {
     char *cp = NULL;
     char const *localedir = relocate2 (LOCALEDIR, &cp);
-    (void) bindtextdomain (PACKAGE, localedir);
-    (void) bindtextdomain ("bison-runtime", localedir);
+    bindtextdomain ("bison", localedir);
+    bindtextdomain ("bison-gnulib", localedir);
+    bindtextdomain ("bison-runtime", localedir);
     free (cp);
   }
-  (void) textdomain (PACKAGE);
+  textdomain ("bison");
 
   {
     char const *cp = getenv ("LC_CTYPE");
@@ -155,7 +156,10 @@ main (int argc, char *argv[])
 
   print_precedence_warnings ();
 
-  if (!update_flag)
+  /* Whether to generate output files.  */
+  bool generate = !(feature_flag & feature_syntax_only);
+
+  if (generate)
     {
       /* Output file names. */
       compute_output_file_names ();
@@ -196,7 +200,7 @@ main (int argc, char *argv[])
   timevar_pop (tv_free);
 
   /* Output the tables and the parser to ftable.  In file output.  */
-  if (!update_flag)
+  if (generate)
     {
       timevar_push (tv_parser);
       output ();
@@ -231,8 +235,6 @@ main (int argc, char *argv[])
   timevar_stop (tv_total);
   timevar_print (stderr);
 
-  cleanup_caret ();
-
   /* Fix input file now, even if there are errors: that's less
      warnings in the following runs.  */
   if (!fixits_empty ())
@@ -246,5 +248,7 @@ main (int argc, char *argv[])
     }
   uniqstrs_free ();
 
+  complain_free ();
+
   return complaint_status ? EXIT_FAILURE : EXIT_SUCCESS;
 }

src/muscle-tab.c

@@ -211,7 +211,7 @@ muscle_syncline_grow (char const *key, location loc)
   obstack_printf (&muscle_obstack, "]b4_syncline(%d, ", loc.start.line);
   obstack_quote (&muscle_obstack,
                  quotearg_style (c_quoting_style, loc.start.file));
-  obstack_sgrow (&muscle_obstack, ")[");
+  obstack_sgrow (&muscle_obstack, ")dnl\n[");
   char const *extension = obstack_finish0 (&muscle_obstack);
   muscle_grow (key, extension, "", "");
   obstack_free (&muscle_obstack, extension);
@@ -227,7 +227,7 @@ void
 muscle_code_grow (const char *key, const char *val, location loc)
 {
   muscle_syncline_grow (key, loc);
-  muscle_grow (key, val, "\n", "\n");
+  muscle_grow (key, val, "", "\n");
 }
 
 
@@ -275,18 +275,14 @@ muscle_boundary_grow (char const *key, boundary bound)
 {
   obstack_sgrow  (&muscle_obstack, "[[");
   obstack_escape (&muscle_obstack, bound.file);
-  obstack_printf (&muscle_obstack, ":%d.%d]]", bound.line, bound.column);
+  obstack_printf (&muscle_obstack, ":%d.%d@@%d]]", bound.line, bound.column, bound.byte);
   char const *extension = obstack_finish0 (&muscle_obstack);
   muscle_grow (key, extension, "", "");
   obstack_free (&muscle_obstack, extension);
 }
 
 
-/* In the format '[[file_name:line.column]], [[file_name:line.column]]',
-   append LOC to MUSCLE.  Use digraphs for special characters in each
-   file name.  */
-
-static void
+void
 muscle_location_grow (char const *key, location loc)
 {
   muscle_boundary_grow (key, loc.start);
@@ -714,9 +710,11 @@ muscle_percent_define_default (char const *variable, char const *value)
       {
         uniqstr loc_name = muscle_name (variable, "loc");
         location loc;
-        loc.start.file = loc.end.file = "<default value>";
-        loc.start.line = loc.end.line = -1;
-        loc.start.column = loc.end.column = -1;
+        loc.start.file = "<default value>";
+        loc.start.line = -1;
+        loc.start.column = -1;
+        loc.start.byte = -1;
+        loc.end = loc.start;
         muscle_insert (loc_name, "");
         muscle_location_grow (loc_name, loc);
       }

src/muscle-tab.h

@@ -119,6 +119,11 @@ void muscle_pair_list_grow (const char *muscle,
 void muscle_user_name_list_grow (char const *key, char const *user_name,
                                  location loc);
 
+/* In the format '[[file_name:line.column]], [[file_name:line.column]]',
+   append LOC to MUSCLE.  Use digraphs for special characters in each
+   file name.  */
+void muscle_location_grow (char const *key, location loc);
+
 /* Indicates whether a variable's value was specified with -D/--define, with
    -F/--force-define, or in the grammar file.  */
 typedef enum {

src/output.c

@@ -21,12 +21,10 @@
 #include <config.h>
 #include "system.h"
 
-#include <configmake.h>
 #include <filename.h> /* IS_PATH_WITH_DIR */
 #include <get-errno.h>
 #include <path-join.h>
 #include <quotearg.h>
-#include <relocatable.h> /* relocate2 */
 #include <spawn-pipe.h>
 #include <timevar.h>
 #include <wait-process.h>
@@ -38,6 +36,7 @@
 #include "muscle-tab.h"
 #include "output.h"
 #include "reader.h"
+#include "reduce.h"
 #include "scan-code.h"    /* max_left_semantic_context */
 #include "scan-skel.h"
 #include "symtab.h"
@@ -45,9 +44,6 @@
 
 static struct obstack format_obstack;
 
-/* Memory allocated by relocate2, to free.  */
-static char *relocate_buffer = NULL;
-
 
 /*-------------------------------------------------------------------.
 | Create a function NAME which associates to the muscle NAME the     |
@@ -137,9 +133,8 @@ string_output (FILE *out, char const *string)
 }
 
 
-/*----------------------------.
-| Prepare the symbols names.  |
-`----------------------------*/
+/* Generate the b4_<MUSCLE_NAME> (e.g., b4_tname) table with the
+   symbol names (aka tags). */
 
 static void
 prepare_symbol_names (char const *muscle_name)
@@ -377,9 +372,9 @@ user_actions_output (FILE *out)
       {
         fprintf (out, "%s(%d, [b4_syncline(%d, ",
                  rules[r].is_predicate ? "b4_predicate_case" : "b4_case",
-                 r + 1, rules[r].action_location.start.line);
-        string_output (out, rules[r].action_location.start.file);
-        fprintf (out, ")\n[    %s]])\n\n", rules[r].action);
+                 r + 1, rules[r].action_loc.start.line);
+        string_output (out, rules[r].action_loc.start.file);
+        fprintf (out, ")dnl\n[    %s]])\n\n", rules[r].action);
       }
   fputs ("])\n\n", out);
 }
@@ -414,6 +409,14 @@ merger_output (FILE *out)
 static void
 prepare_symbol_definitions (void)
 {
+  /* Map "orig NUM" to new numbers.  See data/README.  */
+  for (symbol_number i = ntokens; i < nsyms + nuseless_nonterminals; ++i)
+    {
+      obstack_printf (&format_obstack, "symbol(orig %d, number)", i);
+      const char *key = obstack_finish0 (&format_obstack);
+      MUSCLE_INSERT_INT (key, nterm_map ? nterm_map[i - ntokens] : i);
+    }
+
   for (int i = 0; i < nsyms; ++i)
     {
       symbol *sym = symbols[i];
@@ -430,13 +433,13 @@ prepare_symbol_definitions (void)
       key = obstack_finish0 (&format_obstack);
 
       /* Whether the symbol has an identifier.  */
-      const char *value = symbol_id_get (sym);
+      const char *id = symbol_id_get (sym);
       SET_KEY ("has_id");
-      MUSCLE_INSERT_INT (key, !!value);
+      MUSCLE_INSERT_INT (key, !!id);
 
       /* Its identifier.  */
       SET_KEY ("id");
-      MUSCLE_INSERT_STRING (key, value ? value : "");
+      MUSCLE_INSERT_STRING (key, id ? id : "");
 
       /* Its tag.  Typically for documentation purpose.  */
       SET_KEY ("tag");
@@ -475,6 +478,9 @@ prepare_symbol_definitions (void)
               SET_KEY2 (pname, "line");
               MUSCLE_INSERT_INT (key, p->location.start.line);
 
+              SET_KEY2 (pname, "loc");
+              muscle_location_grow (key, p->location);
+
               SET_KEY (pname);
               MUSCLE_INSERT_STRING_RAW (key, p->code);
             }
@@ -676,7 +682,7 @@ prepare (void)
 #define DEFINE(Name) MUSCLE_INSERT_STRING (#Name, Name ? Name : "")
   DEFINE (dir_prefix);
   DEFINE (parser_file_name);
-  DEFINE (spec_defines_file);
+  DEFINE (spec_header_file);
   DEFINE (spec_file_prefix);
   DEFINE (spec_graph_file);
   DEFINE (spec_name_prefix);
@@ -728,17 +734,4 @@ output (void)
     unlink_generated_sources ();
 
   obstack_free (&format_obstack, NULL);
-  free (relocate_buffer);
-}
-
-char const *
-pkgdatadir (void)
-{
-  if (relocate_buffer)
-    return relocate_buffer;
-  else
-    {
-      char const *cp = getenv ("BISON_PKGDATADIR");
-      return cp ? cp : relocate2 (PKGDATADIR, &relocate_buffer);
-    }
 }

src/output.h

@@ -24,7 +24,4 @@
 /* Output the parsing tables and the parser code to FTABLE.  */
 void output (void);
 
-/* Where our data files are installed.  */
-char const *pkgdatadir (void);
-
 #endif /* !OUTPUT_H_ */

src/parse-gram.h

@@ -1,4 +1,4 @@
-/* A Bison parser, made by GNU Bison 3.2.90.23-0bbcb-dirty.  */
+/* A Bison parser, made by GNU Bison 3.3.2.153-f5a4-dirty.  */
 
 /* Bison interface for Yacc-like parsers in C
 
@@ -52,11 +52,9 @@
 extern int gram_debug;
 #endif
 /* "%code requires" blocks.  */
-#line 21 "src/parse-gram.y" /* yacc.c:1921  */
 
   #include "symlist.h"
   #include "symtab.h"
-#line 239 "src/parse-gram.y" /* yacc.c:1921  */
 
   typedef enum
   {
@@ -65,7 +63,6 @@ extern int gram_debug;
     param_parse  = 1 << 1,
     param_both   = param_lex | param_parse
   } param_type;
-#line 682 "src/parse-gram.y" /* yacc.c:1921  */
 
   #include "muscle-tab.h"
   typedef struct
@@ -74,7 +71,6 @@ extern int gram_debug;
     muscle_kind kind;
   } value_type;
 
-#line 78 "src/parse-gram.h" /* yacc.c:1921  */
 
 /* Token type.  */
 #ifndef GRAM_TOKENTYPE
@@ -122,27 +118,27 @@ extern int gram_debug;
     BRACED_PREDICATE = 295,
     BRACKETED_ID = 296,
     CHAR = 297,
-    EPILOGUE = 298,
-    EQUAL = 299,
-    ID = 300,
-    ID_COLON = 301,
-    PERCENT_PERCENT = 302,
-    PIPE = 303,
-    PROLOGUE = 304,
-    SEMICOLON = 305,
-    TAG = 306,
-    TAG_ANY = 307,
-    TAG_NONE = 308,
-    INT = 309,
-    PERCENT_PARAM = 310,
-    PERCENT_UNION = 311,
-    PERCENT_EMPTY = 312
+    COLON = 298,
+    EPILOGUE = 299,
+    EQUAL = 300,
+    ID = 301,
+    ID_COLON = 302,
+    PERCENT_PERCENT = 303,
+    PIPE = 304,
+    PROLOGUE = 305,
+    SEMICOLON = 306,
+    TAG = 307,
+    TAG_ANY = 308,
+    TAG_NONE = 309,
+    INT = 310,
+    PERCENT_PARAM = 311,
+    PERCENT_UNION = 312,
+    PERCENT_EMPTY = 313
   };
 #endif
 
 /* Value type.  */
 #if ! defined GRAM_STYPE && ! defined GRAM_STYPE_IS_DECLARED
-
 union GRAM_STYPE
 {
 
@@ -163,9 +159,9 @@ union GRAM_STYPE
   /* "integer"  */
   int INT;
   /* int.opt  */
-  int yytype_79;
+  int yytype_80;
   /* named_ref.opt  */
-  named_ref* yytype_91;
+  named_ref* yytype_92;
   /* "%param"  */
   param_type PERCENT_PARAM;
   /* token_decl  */
@@ -181,7 +177,7 @@ union GRAM_STYPE
   /* string_as_id  */
   symbol* string_as_id;
   /* string_as_id.opt  */
-  symbol* yytype_98;
+  symbol* yytype_99;
   /* generic_symlist  */
   symbol_list* generic_symlist;
   /* generic_symlist_item  */
@@ -191,15 +187,15 @@ union GRAM_STYPE
   /* token_decls  */
   symbol_list* token_decls;
   /* token_decl.1  */
-  symbol_list* yytype_77;
+  symbol_list* yytype_78;
   /* token_decls_for_prec  */
   symbol_list* token_decls_for_prec;
   /* token_decl_for_prec.1  */
-  symbol_list* yytype_81;
+  symbol_list* yytype_82;
   /* symbol_decls  */
   symbol_list* symbol_decls;
   /* symbol_decl.1  */
-  symbol_list* yytype_84;
+  symbol_list* yytype_85;
   /* "%error-verbose"  */
   uniqstr PERCENT_ERROR_VERBOSE;
   /* "%<flag>"  */
@@ -219,7 +215,7 @@ union GRAM_STYPE
   /* "<tag>"  */
   uniqstr TAG;
   /* tag.opt  */
-  uniqstr yytype_71;
+  uniqstr yytype_72;
   /* tag  */
   uniqstr tag;
   /* variable  */
@@ -228,9 +224,8 @@ union GRAM_STYPE
   unsigned char CHAR;
   /* value  */
   value_type value;
-#line 232 "src/parse-gram.h" /* yacc.c:1921  */
-};
 
+};
 typedef union GRAM_STYPE GRAM_STYPE;
 # define GRAM_STYPE_IS_TRIVIAL 1
 # define GRAM_STYPE_IS_DECLARED 1

src/parse-gram.y

@@ -49,7 +49,7 @@
   #include "xmemdup0.h"
 
   static int current_prec = 0;
-  static location current_lhs_location;
+  static location current_lhs_loc;
   static named_ref *current_lhs_named_ref;
   static symbol *current_lhs_symbol;
   static symbol_class current_class = unknown_sym;
@@ -112,8 +112,13 @@
   #define YYTYPE_INT8 int_fast8_t
   #define YYTYPE_UINT16 uint_fast16_t
   #define YYTYPE_UINT8 uint_fast8_t
+
+  /* Add style to semantic values in traces.  */
+  static void tron (FILE *yyo);
+  static void troff (FILE *yyo);
 }
 
+%define api.header.include {"parse-gram.h"}
 %define api.prefix {gram_}
 %define api.pure full
 %define api.value.type union
@@ -129,8 +134,8 @@
 {
   /* Bison's grammar can initial empty locations, hence a default
      location is needed. */
-  boundary_set (&@$.start, current_file, 1, 1);
-  boundary_set (&@$.end, current_file, 1, 1);
+  boundary_set (&@$.start, current_file, 1, 1, 1);
+  boundary_set (&@$.end, current_file, 1, 1, 1);
 }
 
 /* Define the tokens together with their human representation.  */
@@ -188,6 +193,7 @@
 %token BRACED_PREDICATE "%?{...}"
 %token BRACKETED_ID    "[identifier]"
 %token CHAR            "char"
+%token COLON           ":"
 %token EPILOGUE        "epilogue"
 %token EQUAL           "="
 %token ID              "identifier"
@@ -200,12 +206,15 @@
 %token TAG_ANY         "<*>"
 %token TAG_NONE        "<>"
 
+ /* Experimental feature, don't rely on it.  */
+%code pre-printer  {tron (yyo);}
+%code post-printer {troff (yyo);}
+
 %type <unsigned char> CHAR
 %printer { fputs (char_name ($$), yyo); } <unsigned char>
 
 %type <char*> "{...}" "%?{...}" "%{...%}" EPILOGUE STRING
-%printer { fputs (quotearg_style (c_quoting_style, $$), yyo); } STRING
-%printer { fprintf (yyo, "{\n%s\n}", $$); } <char*>
+%printer { fputs ($$, yyo); }  <char*>
 
 %type <uniqstr>
   BRACKETED_ID ID ID_COLON
@@ -312,7 +321,7 @@ prologue_declaration:
 | "%defines" STRING
     {
       defines_flag = true;
-      spec_defines_file = xstrdup ($2);
+      spec_header_file = xstrdup ($2);
     }
 | "%error-verbose"                 { handle_error_verbose (&@$, $1); }
 | "%expect" INT                    { expected_sr_conflicts = $2; }
@@ -626,7 +635,7 @@ rules_or_grammar_declaration:
 ;
 
 rules:
-  id_colon named_ref.opt { current_lhs ($1, @1, $2); } rhses.1
+  id_colon named_ref.opt { current_lhs ($1, @1, $2); } ":" rhses.1
     {
       /* Free the current lhs. */
       current_lhs (0, @1, 0);
@@ -634,15 +643,15 @@ rules:
 ;
 
 rhses.1:
-  rhs                { grammar_current_rule_end (@1); }
-| rhses.1 "|" rhs    { grammar_current_rule_end (@3); }
+  rhs                { grammar_current_rule_end (@rhs); }
+| rhses.1 "|" rhs    { grammar_current_rule_end (@rhs); }
 | rhses.1 ";"
 ;
 
 %token PERCENT_EMPTY "%empty";
 rhs:
   %empty
-    { grammar_current_rule_begin (current_lhs_symbol, current_lhs_location,
+    { grammar_current_rule_begin (current_lhs_symbol, current_lhs_loc,
                                   current_lhs_named_ref); }
 | rhs symbol named_ref.opt
     { grammar_current_rule_symbol_append ($2, @2, $3); }
@@ -665,7 +674,7 @@ rhs:
 ;
 
 named_ref.opt:
-  %empty         { $$ = 0; }
+  %empty         { $$ = NULL; }
 | BRACKETED_ID   { $$ = named_ref_new ($1, @1); }
 ;
 
@@ -958,10 +967,9 @@ handle_require (location const *loc, char const *version)
       return;
     }
   required_version = major * 100 + minor;
-  /* Pretend to be at least 3.2, even if we are only 3.1-211, as it
-     allows us to check features published in 3.2 while developping
-     3.2.  */
-  const char* api_version = "3.2";
+  /* Pretend to be at least 3.4, to check features published in 3.4
+     while developping it.  */
+  const char* api_version = "3.4";
   const char* package_version =
     strverscmp (api_version, PACKAGE_VERSION) > 0
     ? api_version : PACKAGE_VERSION;
@@ -1042,7 +1050,7 @@ void
 current_lhs (symbol *sym, location loc, named_ref *ref)
 {
   current_lhs_symbol = sym;
-  current_lhs_location = loc;
+  current_lhs_loc = loc;
   if (sym)
     symbol_location_as_lhs_set (sym, loc);
   /* In order to simplify memory management, named references for lhs
@@ -1053,3 +1061,13 @@ current_lhs (symbol *sym, location loc, named_ref *ref)
   free (current_lhs_named_ref);
   current_lhs_named_ref = ref;
 }
+
+static void tron (FILE *yyo)
+{
+  begin_use_class ("value", yyo);
+}
+
+static void troff (FILE *yyo)
+{
+  end_use_class ("value", yyo);
+}

src/print-graph.c

@@ -19,9 +19,10 @@
    along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
 
 #include <config.h>
+#include "print-graph.h"
+
 #include "system.h"
 
-#include "LR0.h"
 #include "closure.h"
 #include "complain.h"
 #include "conflicts.h"
@@ -30,7 +31,7 @@
 #include "gram.h"
 #include "graphviz.h"
 #include "lalr.h"
-#include "print_graph.h"
+#include "lr0.h"
 #include "reader.h"
 #include "state.h"
 #include "symtab.h"
@@ -63,12 +64,7 @@ print_core (struct obstack *oout, state *s)
   for (size_t i = 0; i < snritems; ++i)
     {
       item_number const *sp1 = ritem + sitems[i];
-      item_number const *sp = sp1;
-
-      while (0 <= *sp)
-        sp++;
-
-      rule *r = &rules[item_number_as_rule_number (*sp)];
+      rule const *r = item_rule (sp1);
 
       obstack_printf (oout, "%3d ", r->number);
       if (previous_lhs && UNIQSTR_EQ (previous_lhs->symbol->tag,
@@ -79,13 +75,13 @@ print_core (struct obstack *oout, state *s)
         obstack_printf (oout, "%s: ", escape (r->lhs->symbol->tag));
       previous_lhs = r->lhs;
 
-      for (sp = r->rhs; sp < sp1; sp++)
+      for (item_number const *sp = r->rhs; sp < sp1; sp++)
         obstack_printf (oout, "%s ", escape (symbols[*sp]->tag));
 
       obstack_1grow (oout, '.');
 
       if (0 <= *r->rhs)
-        for (/* Nothing */; *sp >= 0; ++sp)
+        for (item_number const *sp = sp1; 0 <= *sp; ++sp)
           obstack_printf (oout, " %s", escape (symbols[*sp]->tag));
       else
         obstack_printf (oout, " %%empty");
@@ -184,10 +180,8 @@ print_graph (void)
   start_graph (fgraph);
 
   /* Output nodes and edges. */
-  new_closure (nritems);
   for (int i = 0; i < nstates; i++)
     print_state (states[i], fgraph);
-  free_closure ();
 
   finish_graph (fgraph);
   xfclose (fgraph);

src/print-xml.c

@@ -19,21 +19,21 @@
    along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
 
 #include <config.h>
+#include "print-xml.h"
+
 #include "system.h"
 
+#include <bitset.h>
 #include <stdarg.h>
 
-#include <bitset.h>
-
-#include "LR0.h"
 #include "closure.h"
 #include "conflicts.h"
 #include "files.h"
 #include "getargs.h"
 #include "gram.h"
 #include "lalr.h"
+#include "lr0.h"
 #include "print.h"
-#include "print-xml.h"
 #include "reader.h"
 #include "reduce.h"
 #include "state.h"
@@ -57,7 +57,6 @@ static struct escape_buf escape_bufs[num_escape_bufs];
 static void
 print_core (FILE *out, int level, state *s)
 {
-  size_t i;
   item_number *sitems = s->items;
   size_t snritems = s->nitems;
 
@@ -74,18 +73,12 @@ print_core (FILE *out, int level, state *s)
 
   xml_puts (out, level, "<itemset>");
 
-  for (i = 0; i < snritems; i++)
+  for (size_t i = 0; i < snritems; i++)
     {
       bool printed = false;
       item_number *sp1 = ritem + sitems[i];
-      item_number *sp = sp1;
-      rule *r;
-
-      while (0 <= *sp)
-        sp++;
-
-      r = &rules[item_number_as_rule_number (*sp)];
-      sp = r->rhs;
+      rule const *r = item_rule (sp1);
+      item_number *sp = r->rhs;
 
       /* Display the lookahead tokens?  */
       if (item_number_is_rule_number (*sp1))
@@ -380,15 +373,13 @@ print_state (FILE *out, int level, state *s)
 static void
 print_grammar (FILE *out, int level)
 {
-  symbol_number i;
-
   fputc ('\n', out);
   xml_puts (out, level, "<grammar>");
   grammar_rules_print_xml (out, level);
 
   /* Terminals */
   xml_puts (out, level + 1, "<terminals>");
-  for (i = 0; i < max_user_token_number + 1; i++)
+  for (symbol_number i = 0; i < max_user_token_number + 1; i++)
     if (token_translations[i] != undeftoken->content->number)
       {
         char const *tag = symbols[token_translations[i]]->tag;
@@ -411,7 +402,7 @@ print_grammar (FILE *out, int level)
 
   /* Nonterminals */
   xml_puts (out, level + 1, "<nonterminals>");
-  for (i = ntokens; i < nsyms + nuseless_nonterminals; i++)
+  for (symbol_number i = ntokens; i < nsyms + nuseless_nonterminals; i++)
     {
       char const *tag = symbols[i]->tag;
       xml_printf (out, level + 2,
@@ -428,8 +419,7 @@ print_grammar (FILE *out, int level)
 void
 xml_indent (FILE *out, int level)
 {
-  int i;
-  for (i = 0; i < level; i++)
+  for (int i = 0; i < level; i++)
     fputs ("  ", out);
 }
 
@@ -460,14 +450,13 @@ xml_escape_string (struct escape_buf *buf, char const *str)
 {
   size_t len = strlen (str);
   size_t max_expansion = sizeof "&quot;" - 1;
-  char *p;
 
   if (buf->size <= max_expansion * len)
     {
       buf->size = max_expansion * len + 1;
       buf->ptr = x2realloc (buf->ptr, &buf->size);
     }
-  p = buf->ptr;
+  char *p = buf->ptr;
 
   for (; *str; str++)
     switch (*str)
@@ -498,11 +487,11 @@ xml_escape (char const *str)
 void
 print_xml (void)
 {
-  int level = 0;
-
   FILE *out = xfopen (spec_xml_file, "w");
 
   fputs ("<?xml version=\"1.0\"?>\n\n", out);
+
+  int level = 0;
   xml_printf (out, level,
               "<bison-xml-report version=\"%s\" bug-report=\"%s\""
               " url=\"%s\">",
@@ -517,29 +506,21 @@ print_xml (void)
   /* print grammar */
   print_grammar (out, level + 1);
 
-  new_closure (nritems);
-  no_reduce_set =  bitset_create (ntokens, BITSET_FIXED);
+  no_reduce_set = bitset_create (ntokens, BITSET_FIXED);
 
   /* print automaton */
   fputc ('\n', out);
   xml_puts (out, level + 1, "<automaton>");
-  {
-    state_number i;
-    for (i = 0; i < nstates; i++)
-      print_state (out, level + 2, states[i]);
-  }
+  for (state_number i = 0; i < nstates; i++)
+    print_state (out, level + 2, states[i]);
   xml_puts (out, level + 1, "</automaton>");
 
   bitset_free (no_reduce_set);
-  free_closure ();
 
   xml_puts (out, 0, "</bison-xml-report>");
 
-  {
-    int i;
-    for (i = 0; i < num_escape_bufs; ++i)
-      free (escape_bufs[i].ptr);
-  }
+  for (int i = 0; i < num_escape_bufs; ++i)
+    free (escape_bufs[i].ptr);
 
   xfclose (out);
 }

src/print-xml.h

@@ -21,6 +21,10 @@
 #ifndef PRINT_XML_H_
 # define PRINT_XML_H_
 
+# include <config.h>
+
+# include <stdio.h>
+
 void xml_indent (FILE *out, int level);
 void xml_puts (FILE *, int, char const *);
 void xml_printf (FILE *, int, char const *, ...);

src/print.c

@@ -23,13 +23,13 @@
 
 #include <bitset.h>
 
-#include "LR0.h"
 #include "closure.h"
 #include "conflicts.h"
 #include "files.h"
 #include "getargs.h"
 #include "gram.h"
 #include "lalr.h"
+#include "lr0.h"
 #include "muscle-tab.h"
 #include "print.h"
 #include "reader.h"
@@ -63,8 +63,6 @@ print_core (FILE *out, state *s)
 {
   item_number *sitems = s->items;
   size_t snritems = s->nitems;
-  sym_content *previous_lhs = NULL;
-
   /* Output all the items of a state, not only its kernel.  */
   if (report_flag & report_itemsets)
     {
@@ -78,33 +76,18 @@ print_core (FILE *out, state *s)
 
   fputc ('\n', out);
 
+  rule const *previous_rule = NULL;
   for (size_t i = 0; i < snritems; i++)
     {
       item_number *sp1 = ritem + sitems[i];
-
-      item_number *sp = sp1;
-      while (*sp >= 0)
-        sp++;
-
-      rule_number r = item_number_as_rule_number (*sp);
-
-      rule_lhs_print (&rules[r], previous_lhs, out);
-      previous_lhs = rules[r].lhs;
-
-      for (sp = rules[r].rhs; sp < sp1; sp++)
-        fprintf (out, " %s", symbols[*sp]->tag);
-      fputs (" .", out);
-      if (0 <= *rules[r].rhs)
-        for (/* Nothing */; 0 <= *sp; ++sp)
-          fprintf (out, " %s", symbols[*sp]->tag);
-      else
-        fprintf (out, " %%empty");
+      rule const *r = item_rule (sp1);
+      item_print (sp1, previous_rule, out);
+      previous_rule = r;
 
       /* Display the lookahead tokens?  */
       if (report_flag & report_lookahead_tokens
           && item_number_is_rule_number (*sp1))
-        state_rule_lookahead_tokens_print (s, &rules[r], out);
-
+        state_rule_lookahead_tokens_print (s, r, out);
       fputc ('\n', out);
     }
 }
@@ -374,17 +357,6 @@ print_state (FILE *out, state *s)
 | Print information on the whole grammar.  |
 `-----------------------------------------*/
 
-#define END_TEST(End)                           \
-  do {                                          \
-    if (column + strlen (buffer) > (End))       \
-      {                                         \
-        fprintf (out, "%s\n   ", buffer);       \
-        column = 3;                             \
-        buffer[0] = 0;                          \
-      }                                         \
-  } while (0)
-
-
 static void
 print_terminal_symbols (FILE *out)
 {
@@ -394,29 +366,20 @@ print_terminal_symbols (FILE *out)
     if (token_translations[i] != undeftoken->content->number)
       {
         const char *tag = symbols[token_translations[i]]->tag;
-        int column = strlen (tag);
-        char buffer[90];
-
-        buffer[0] = 0;
-        fputs (tag, out);
-        END_TEST (65);
+        fprintf (out, "%4s%s", "", tag);
         if (symbols[token_translations[i]]->content->type_name)
-          {
-            column += fprintf (out, " <%s>",
-                               symbols[token_translations[i]]->content->type_name);
-            END_TEST (65);
-          }
-        sprintf (buffer, " (%d)", i);
+          fprintf (out, " <%s>",
+                   symbols[token_translations[i]]->content->type_name);
+        fprintf (out, " (%d)", i);
 
         for (rule_number r = 0; r < nrules; r++)
           for (item_number *rhsp = rules[r].rhs; *rhsp >= 0; rhsp++)
             if (item_number_as_symbol_number (*rhsp) == token_translations[i])
               {
-                END_TEST (65);
-                sprintf (buffer + strlen (buffer), " %d", r);
+                fprintf (out, " %d", r);
                 break;
               }
-        fprintf (out, "%s\n", buffer);
+        fputc ('\n', out);
       }
   fputs ("\n\n", out);
 }
@@ -429,64 +392,46 @@ print_nonterminal_symbols (FILE *out)
   for (symbol_number i = ntokens; i < nsyms; i++)
     {
       const char *tag = symbols[i]->tag;
-      int column = strlen (tag);
-      int left_count = 0, right_count = 0;
+      bool on_left = false;
+      bool on_right = false;
 
       for (rule_number r = 0; r < nrules; r++)
         {
-          if (rules[r].lhs->number == i)
-            left_count++;
-          for (item_number *rhsp = rules[r].rhs; *rhsp >= 0; rhsp++)
-            if (item_number_as_symbol_number (*rhsp) == i)
-              {
-                right_count++;
-                break;
-              }
+          on_left |= rules[r].lhs->number == i;
+          for (item_number *rhsp = rules[r].rhs; !on_right && 0 <= *rhsp; ++rhsp)
+            on_right |= item_number_as_symbol_number (*rhsp) == i;
+          if (on_left && on_right)
+            break;
         }
 
-      fputs (tag, out);
+      int column = 4 + strlen (tag);
+      fprintf (out, "%4s%s", "", tag);
       if (symbols[i]->content->type_name)
         column += fprintf (out, " <%s>",
                            symbols[i]->content->type_name);
-      char buffer[90];
-      buffer[0] = 0;
-      sprintf (buffer, " (%d)", i);
-      END_TEST (0);
+      fprintf (out, " (%d)\n", i);
 
-      if (left_count > 0)
+      if (on_left)
         {
-          END_TEST (65);
-          sprintf (buffer + strlen (buffer), _(" on left:"));
-
+          fprintf (out, "%8s%s", "", _("on left:"));
           for (rule_number r = 0; r < nrules; r++)
-            {
-              if (rules[r].lhs->number == i)
+            if (rules[r].lhs->number == i)
+              fprintf (out, " %d", r);
+          fputc ('\n', out);
+        }
+
+      if (on_right)
+        {
+          fprintf (out, "%8s%s", "", _("on right:"));
+          for (rule_number r = 0; r < nrules; r++)
+            for (item_number *rhsp = rules[r].rhs; 0 <= *rhsp; ++rhsp)
+              if (item_number_as_symbol_number (*rhsp) == i)
                 {
-                  END_TEST (65);
-                  sprintf (buffer + strlen (buffer), " %d", r);
+                  fprintf (out, " %d", r);
+                  break;
                 }
-            }
+          fputc ('\n', out);
         }
-
-      if (right_count > 0)
-        {
-          if (left_count > 0)
-            sprintf (buffer + strlen (buffer), ",");
-          END_TEST (65);
-          sprintf (buffer + strlen (buffer), _(" on right:"));
-          for (rule_number r = 0; r < nrules; r++)
-            {
-              item_number *rhsp;
-              for (rhsp = rules[r].rhs; *rhsp >= 0; rhsp++)
-                if (item_number_as_symbol_number (*rhsp) == i)
-                  {
-                    END_TEST (65);
-                    sprintf (buffer + strlen (buffer), " %d", r);
-                    break;
-                  }
-            }
-        }
-      fprintf (out, "%s\n", buffer);
     }
 }
 
@@ -507,17 +452,11 @@ print_results (void)
   print_terminal_symbols (out);
   print_nonterminal_symbols (out);
 
-  /* If the whole state item sets, not only the kernels, are wanted,
-     'closure' will be run, which needs memory allocation/deallocation.   */
-  if (report_flag & report_itemsets)
-    new_closure (nritems);
   /* Storage for print_reductions.  */
   no_reduce_set = bitset_create (ntokens, BITSET_FIXED);
   for (state_number i = 0; i < nstates; i++)
     print_state (out, states[i]);
   bitset_free (no_reduce_set);
-  if (report_flag & report_itemsets)
-    free_closure ();
 
   xfclose (out);
 }

src/reader.c

@@ -26,6 +26,7 @@
 #include "complain.h"
 #include "conflicts.h"
 #include "files.h"
+#include "fixits.h"
 #include "getargs.h"
 #include "gram.h"
 #include "muscle-tab.h"
@@ -61,7 +62,7 @@ grammar_start_symbol_set (symbol *sym, location loc)
     {
       start_flag = true;
       startsymbol = sym;
-      startsymbol_location = loc;
+      startsymbol_loc = loc;
     }
 }
 
@@ -130,12 +131,12 @@ record_merge_function_type (int merger, uniqstr type, location declaration_loc)
                        quote (merge_function->name), type,
                        merge_function->type);
       indent += SUB_INDENT;
-      complain_indent (&merge_function->type_declaration_location, complaint,
+      complain_indent (&merge_function->type_declaration_loc, complaint,
                        &indent,
                        _("previous declaration"));
     }
   merge_function->type = uniqstr_new (type);
-  merge_function->type_declaration_location = declaration_loc;
+  merge_function->type_declaration_loc = declaration_loc;
 }
 
 /*--------------------------------------.
@@ -226,11 +227,9 @@ grammar_current_rule_begin (symbol *lhs, location loc,
   ++nrules;
   previous_rule_end = grammar_end;
 
-  symbol_list *p = grammar_symbol_append (lhs, loc);
+  current_rule = grammar_symbol_append (lhs, loc);
   if (lhs_name)
-    assign_named_ref (p, named_ref_copy (lhs_name));
-
-  current_rule = grammar_end;
+    assign_named_ref (current_rule, named_ref_copy (lhs_name));
 
   /* Mark the rule's lhs as a nonterminal if not already so.  */
   if (lhs->content->class == unknown_sym)
@@ -293,10 +292,10 @@ grammar_rule_check_and_complete (symbol_list *r)
       if (first_rhs)
         {
           char const *lhs_type = r->content.sym->content->type_name;
-          const char *rhs_type =
+          char const *rhs_type =
             first_rhs->content->type_name ? first_rhs->content->type_name : "";
           if (!UNIQSTR_EQ (lhs_type, rhs_type))
-            complain (&r->location, Wother,
+            complain (&r->rhs_loc, Wother,
                       _("type clash on default action: <%s> != <%s>"),
                       lhs_type, rhs_type);
           else
@@ -309,7 +308,7 @@ grammar_rule_check_and_complete (symbol_list *r)
               if (is_cxx)
                 {
                   code_props_rule_action_init (&r->action_props, "{ $$ = $1; }",
-                                               r->location, r,
+                                               r->rhs_loc, r,
                                                /* name */ NULL,
                                                /* type */ NULL,
                                                /* is_predicate */ false);
@@ -319,7 +318,7 @@ grammar_rule_check_and_complete (symbol_list *r)
         }
       /* Warn if there is no default for $$ but we need one.  */
       else
-        complain (&r->location, Wother,
+        complain (&r->rhs_loc, Wother,
                   _("empty rule for typed nonterminal, and no action"));
     }
 
@@ -336,9 +335,9 @@ grammar_rule_check_and_complete (symbol_list *r)
           {
             warnings warn_flag = midrule_warning ? Wmidrule_values : Wother;
             if (n)
-              complain (&l->location, warn_flag, _("unused value: $%d"), n);
+              complain (&l->sym_loc, warn_flag, _("unused value: $%d"), n);
             else
-              complain (&l->location, warn_flag, _("unset value: $$"));
+              complain (&l->rhs_loc, warn_flag, _("unset value: $$"));
           }
       }
   }
@@ -346,14 +345,23 @@ grammar_rule_check_and_complete (symbol_list *r)
   /* Check that %empty => empty rule.  */
   if (r->percent_empty_loc.start.file
       && r->next && r->next->content.sym)
-    complain (&r->percent_empty_loc, complaint,
-              _("%%empty on non-empty rule"));
+    {
+      complain (&r->percent_empty_loc, complaint,
+                _("%%empty on non-empty rule"));
+      fixits_register (&r->percent_empty_loc, "");
+    }
 
   /* Check that empty rule => %empty.  */
   if (!(r->next && r->next->content.sym)
       && !r->midrule_parent_rule
-      && !r->percent_empty_loc.start.file)
-    complain (&r->location, Wempty_rule, _("empty rule without %%empty"));
+      && !r->percent_empty_loc.start.file
+      && warning_is_enabled (Wempty_rule))
+    {
+      complain (&r->rhs_loc, Wempty_rule, _("empty rule without %%empty"));
+      location loc = r->rhs_loc;
+      loc.end = loc.start;
+      fixits_register (&loc, " %empty ");
+    }
 
   /* See comments in grammar_current_rule_prec_set for how POSIX
      mandates this complaint.  It's only for identifiers, so skip
@@ -362,12 +370,12 @@ grammar_rule_check_and_complete (symbol_list *r)
       && r->ruleprec->tag[0] != '\'' && r->ruleprec->tag[0] != '"'
       && r->ruleprec->content->status != declared
       && !r->ruleprec->content->prec)
-    complain (&r->location, Wother,
+    complain (&r->rhs_loc, Wother,
               _("token for %%prec is not defined: %s"), r->ruleprec->tag);
 
   /* Check that the (main) action was not typed.  */
   if (r->action_props.type)
-    complain (&r->location, Wother,
+    complain (&r->rhs_loc, Wother,
               _("only midrule actions can be typed: %s"), r->action_props.type);
 }
 
@@ -380,8 +388,8 @@ void
 grammar_current_rule_end (location loc)
 {
   /* Put an empty link in the list to mark the end of this rule  */
-  grammar_symbol_append (NULL, grammar_end->location);
-  current_rule->location = loc;
+  grammar_symbol_append (NULL, grammar_end->rhs_loc);
+  current_rule->rhs_loc = loc;
 }
 
 
@@ -401,11 +409,11 @@ grammar_midrule_action (void)
 
   /* Make a DUMMY nonterminal, whose location is that of the midrule
      action.  Create the MIDRULE.  */
-  location dummy_location = current_rule->action_props.location;
-  symbol *dummy = dummy_symbol_get (dummy_location);
+  location dummy_loc = current_rule->action_props.location;
+  symbol *dummy = dummy_symbol_get (dummy_loc);
   symbol_type_set(dummy,
                   current_rule->action_props.type, current_rule->action_props.location);
-  symbol_list *midrule = symbol_list_sym_new (dummy, dummy_location);
+  symbol_list *midrule = symbol_list_sym_new (dummy, dummy_loc);
 
   /* Remember named_ref of previous action. */
   named_ref *action_name = current_rule->action_props.named_ref;
@@ -415,7 +423,7 @@ grammar_midrule_action (void)
   ++nrules;
   ++nritems;
   /* Attach its location and actions to that of the DUMMY.  */
-  midrule->location = dummy_location;
+  midrule->rhs_loc = dummy_loc;
   code_props_rule_action_init (&midrule->action_props,
                                current_rule->action_props.code,
                                current_rule->action_props.location,
@@ -436,14 +444,14 @@ grammar_midrule_action (void)
     grammar = midrule;
 
   /* End the dummy's rule.  */
-  midrule->next = symbol_list_sym_new (NULL, dummy_location);
+  midrule->next = symbol_list_sym_new (NULL, dummy_loc);
   midrule->next->next = current_rule;
 
   previous_rule_end = midrule->next;
 
   /* Insert the dummy nonterminal replacing the midrule action into
      the current rule.  Bind it to its dedicated rule.  */
-  grammar_current_rule_symbol_append (dummy, dummy_location,
+  grammar_current_rule_symbol_append (dummy, dummy_loc,
                                       action_name);
   grammar_end->midrule = midrule;
   midrule->midrule_parent_rule = current_rule;
@@ -502,11 +510,11 @@ grammar_current_rule_dprec_set (int dprec, location loc)
               "%dprec");
   else if (current_rule->dprec != 0)
     duplicate_rule_directive ("%dprec",
-                              current_rule->dprec_location, loc);
+                              current_rule->dprec_loc, loc);
   else
     {
       current_rule->dprec = dprec;
-      current_rule->dprec_location = loc;
+      current_rule->dprec_loc = loc;
     }
 }
 
@@ -521,11 +529,11 @@ grammar_current_rule_merge_set (uniqstr name, location loc)
               "%merge");
   if (current_rule->merger != 0)
     duplicate_rule_directive ("%merge",
-                              current_rule->merger_declaration_location, loc);
+                              current_rule->merger_declaration_loc, loc);
   else
     {
       current_rule->merger = get_merge_function (name);
-      current_rule->merger_declaration_location = loc;
+      current_rule->merger_declaration_loc = loc;
     }
 }
 
@@ -615,11 +623,11 @@ packgram (void)
     {
       symbol_list *lhs = p;
       record_merge_function_type (lhs->merger, lhs->content.sym->content->type_name,
-                                  lhs->merger_declaration_location);
+                                  lhs->merger_declaration_loc);
       /* If the midrule's $$ is set or its $n is used, remove the '$' from the
          symbol name so that it's a user-defined symbol so that the default
          %destructor and %printer apply.  */
-      if (lhs->midrule_parent_rule
+      if (lhs->midrule_parent_rule /* i.e., symbol_is_dummy (lhs->content.sym).  */
           && (lhs->action_props.is_value_used
               || (symbol_list_n_get (lhs->midrule_parent_rule,
                                      lhs->midrule_parent_rhs_index)
@@ -629,8 +637,8 @@ packgram (void)
       /* Don't check the generated rule 0.  It has no action, so some rhs
          symbols may appear unused, but the parsing algorithm ensures that
          %destructor's are invoked appropriately.  */
-      if (p != grammar)
-        grammar_rule_check_and_complete (p);
+      if (lhs != grammar)
+        grammar_rule_check_and_complete (lhs);
 
       rules[ruleno].user_number = ruleno;
       rules[ruleno].number = ruleno;
@@ -640,18 +648,18 @@ packgram (void)
       rules[ruleno].dprec = lhs->dprec;
       rules[ruleno].merger = lhs->merger;
       rules[ruleno].precsym = NULL;
-      rules[ruleno].location = lhs->location;
+      rules[ruleno].location = lhs->rhs_loc;
       rules[ruleno].useful = true;
       rules[ruleno].action = lhs->action_props.code;
-      rules[ruleno].action_location = lhs->action_props.location;
+      rules[ruleno].action_loc = lhs->action_props.location;
       rules[ruleno].is_predicate = lhs->action_props.is_predicate;
-      rules[ruleno].expected_sr_conflicts = p->expected_sr_conflicts;
-      rules[ruleno].expected_rr_conflicts = p->expected_rr_conflicts;
+      rules[ruleno].expected_sr_conflicts = lhs->expected_sr_conflicts;
+      rules[ruleno].expected_rr_conflicts = lhs->expected_rr_conflicts;
 
       /* Traverse the rhs.  */
       {
         size_t rule_length = 0;
-        for (p = p->next; p->content.sym; p = p->next)
+        for (p = lhs->next; p->content.sym; p = p->next)
           {
             ++rule_length;
 
@@ -706,18 +714,18 @@ reader (void)
   symbols_new ();
 
   /* Construct the accept symbol. */
-  accept = symbol_get ("$accept", empty_location);
+  accept = symbol_get ("$accept", empty_loc);
   accept->content->class = nterm_sym;
   accept->content->number = nvars++;
 
   /* Construct the error token */
-  errtoken = symbol_get ("error", empty_location);
+  errtoken = symbol_get ("error", empty_loc);
   errtoken->content->class = token_sym;
   errtoken->content->number = ntokens++;
 
   /* Construct a token that represents all undefined literal tokens.
      It is always token number 2.  */
-  undeftoken = symbol_get ("$undefined", empty_location);
+  undeftoken = symbol_get ("$undefined", empty_loc);
   undeftoken->content->class = token_sym;
   undeftoken->content->number = ntokens++;
 
@@ -754,11 +762,12 @@ prepare_percent_define_front_end_variables (void)
 
   /* Check %define front-end variables.  */
   {
-    static char const * const values[] = {
-      "lr.type", "lalr", "ielr", "canonical-lr", NULL,
-      "lr.default-reduction", "most", "consistent", "accepting", NULL,
-      NULL
-    };
+    static char const * const values[] =
+      {
+       "lr.type", "lr""(0)", "lalr", "ielr", "canonical-lr", NULL,
+       "lr.default-reduction", "most", "consistent", "accepting", NULL,
+       NULL
+      };
     muscle_percent_define_check_values (values);
   }
 }
@@ -793,7 +802,7 @@ check_and_convert_grammar (void)
   /* If the user did not define her ENDTOKEN, do it now. */
   if (!endtoken)
     {
-      endtoken = symbol_get ("$end", empty_location);
+      endtoken = symbol_get ("$end", empty_loc);
       endtoken->content->class = token_sym;
       endtoken->content->number = 0;
       /* Value specified by POSIX.  */
@@ -815,11 +824,11 @@ check_and_convert_grammar (void)
 
      $accept: %start $end.  */
   {
-    symbol_list *p = symbol_list_sym_new (accept, empty_location);
-    p->location = grammar->location;
-    p->next = symbol_list_sym_new (startsymbol, empty_location);
-    p->next->next = symbol_list_sym_new (endtoken, empty_location);
-    p->next->next->next = symbol_list_sym_new (NULL, empty_location);
+    symbol_list *p = symbol_list_sym_new (accept, empty_loc);
+    p->rhs_loc = grammar->rhs_loc;
+    p->next = symbol_list_sym_new (startsymbol, empty_loc);
+    p->next->next = symbol_list_sym_new (endtoken, empty_loc);
+    p->next->next->next = symbol_list_sym_new (NULL, empty_loc);
     p->next->next->next->next = grammar;
     nrules += 1;
     nritems += 3;
@@ -829,8 +838,7 @@ check_and_convert_grammar (void)
   aver (nsyms <= SYMBOL_NUMBER_MAXIMUM);
   aver (nsyms == ntokens + nvars);
 
-  /* Assign the symbols their symbol numbers.  Write #defines for the
-     token symbols into FDEFINES if requested.  */
+  /* Assign the symbols their symbol numbers.  */
   symbols_pack ();
 
   /* Scan rule actions after invoking symbol_check_alias_consistency (in

src/reader.h

@@ -32,7 +32,7 @@ typedef struct merger_list
   struct merger_list* next;
   uniqstr name;
   uniqstr type;
-  location type_declaration_location;
+  location type_declaration_loc;
 } merger_list;
 
 /* From the parser.  */

src/reduce.c

@@ -39,13 +39,13 @@
 #include "reduce.h"
 #include "symtab.h"
 
-/* Set of all nonterminals whose language is not empty.  */
+/* Set of nonterminals whose language is not empty.  */
 static bitset N;
 
-/* Set of all rules which have no useless nonterminals in their RHS.  */
+/* Set of rules that have no useless nonterminals in their RHS.  */
 static bitset P;
 
-/* Set of all accessible symbols.  */
+/* Set of accessible symbols.  */
 static bitset V;
 
 /* Set of symbols used to define rule precedence (so they are
@@ -81,9 +81,11 @@ useful_production (rule_number r, bitset N0)
 }
 
 
-/*---------------------------------------------------------.
-| Remember that rules are 1-origin, symbols are 0-origin.  |
-`---------------------------------------------------------*/
+/*-----------------------------------------------------------------.
+| Compute N, the set of nonterminals whose language is not empty.  |
+|                                                                  |
+| Remember that rules are 1-origin, symbols are 0-origin.          |
+`-----------------------------------------------------------------*/
 
 static void
 useless_nonterminals (void)
@@ -248,10 +250,7 @@ reduce_grammar_tables (void)
 
   /* Adjust NRITEMS.  */
   for (rule_number r = nrules; r < nrules + nuseless_productions; ++r)
-    {
-      int length = rule_rhs_length (&rules[r]);
-      nritems -= length + 1;
-  }
+    nritems -= rule_rhs_length (&rules[r]) + 1;
 }
 
 
@@ -259,22 +258,23 @@ reduce_grammar_tables (void)
 | Remove useless nonterminals.  |
 `------------------------------*/
 
+symbol_number *nterm_map = NULL;
+
 static void
 nonterminals_reduce (void)
 {
+  nterm_map = xnmalloc (nvars, sizeof *nterm_map);
   /* Map the nonterminals to their new index: useful first, useless
      afterwards.  Kept for later report.  */
-
-  symbol_number *nontermmap = xnmalloc (nvars, sizeof *nontermmap);
   {
     symbol_number n = ntokens;
     for (symbol_number i = ntokens; i < nsyms; ++i)
       if (bitset_test (V, i))
-        nontermmap[i - ntokens] = n++;
+        nterm_map[i - ntokens] = n++;
     for (symbol_number i = ntokens; i < nsyms; ++i)
       if (!bitset_test (V, i))
         {
-          nontermmap[i - ntokens] = n++;
+          nterm_map[i - ntokens] = n++;
           if (symbols[i]->content->status != used)
             complain (&symbols[i]->location, Wother,
                       _("nonterminal useless in grammar: %s"),
@@ -282,32 +282,30 @@ nonterminals_reduce (void)
         }
   }
 
-
   /* Shuffle elements of tables indexed by symbol number.  */
   {
     symbol **symbols_sorted = xnmalloc (nvars, sizeof *symbols_sorted);
     for (symbol_number i = ntokens; i < nsyms; ++i)
-      symbols[i]->content->number = nontermmap[i - ntokens];
+      symbols[i]->content->number = nterm_map[i - ntokens];
     for (symbol_number i = ntokens; i < nsyms; ++i)
-      symbols_sorted[nontermmap[i - ntokens] - ntokens] = symbols[i];
+      symbols_sorted[nterm_map[i - ntokens] - ntokens] = symbols[i];
     for (symbol_number i = ntokens; i < nsyms; ++i)
       symbols[i] = symbols_sorted[i - ntokens];
     free (symbols_sorted);
   }
 
+  /* Update nonterminal numbers in the RHS of the rules.  LHS are
+     pointers to the symbol structure, they don't need renumbering. */
   {
     for (rule_number r = 0; r < nrules; ++r)
       for (item_number *rhsp = rules[r].rhs; 0 <= *rhsp; ++rhsp)
         if (ISVAR (*rhsp))
-          *rhsp =  symbol_number_as_item_number (nontermmap[*rhsp
-                                                            - ntokens]);
-    accept->content->number = nontermmap[accept->content->number - ntokens];
+          *rhsp = symbol_number_as_item_number (nterm_map[*rhsp - ntokens]);
+    accept->content->number = nterm_map[accept->content->number - ntokens];
   }
 
   nsyms -= nuseless_nonterminals;
   nvars -= nuseless_nonterminals;
-
-  free (nontermmap);
 }
 
 
@@ -322,7 +320,7 @@ reduce_output (FILE *out)
     {
       fprintf (out, "%s\n\n", _("Nonterminals useless in grammar"));
       for (int i = 0; i < nuseless_nonterminals; ++i)
-        fprintf (out, "   %s\n", symbols[nsyms + i]->tag);
+        fprintf (out, "    %s\n", symbols[nsyms + i]->tag);
       fputs ("\n\n", out);
     }
 
@@ -334,7 +332,7 @@ reduce_output (FILE *out)
           if (!b)
             fprintf (out, "%s\n\n", _("Terminals unused in grammar"));
           b = true;
-          fprintf (out, "   %s\n", symbols[i]->tag);
+          fprintf (out, "    %s\n", symbols[i]->tag);
         }
     if (b)
       fputs ("\n\n", out);
@@ -379,23 +377,23 @@ reduce_grammar (void)
   inaccessable_symbols ();
 
   /* Did we reduce something? */
-  if (!nuseless_nonterminals && !nuseless_productions)
-    return;
+  if (nuseless_nonterminals || nuseless_productions)
+    {
+      reduce_print ();
 
-  reduce_print ();
+      if (!bitset_test (N, accept->content->number - ntokens))
+        complain (&startsymbol_loc, fatal,
+                  _("start symbol %s does not derive any sentence"),
+                  startsymbol->tag);
 
-  if (!bitset_test (N, accept->content->number - ntokens))
-    complain (&startsymbol_location, fatal,
-              _("start symbol %s does not derive any sentence"),
-              startsymbol->tag);
-
-  /* First reduce the nonterminals, as they renumber themselves in the
-     whole grammar.  If you change the order, nonterms would be
-     renumbered only in the reduced grammar.  */
-  if (nuseless_nonterminals)
-    nonterminals_reduce ();
-  if (nuseless_productions)
-    reduce_grammar_tables ();
+      /* First reduce the nonterminals, as they renumber themselves in the
+         whole grammar.  If you change the order, nonterms would be
+         renumbered only in the reduced grammar.  */
+      if (nuseless_nonterminals)
+        nonterminals_reduce ();
+      if (nuseless_productions)
+        reduce_grammar_tables ();
+    }
 
   if (trace_flag & trace_grammar)
     {
@@ -433,4 +431,6 @@ reduce_free (void)
   bitset_free (V);
   bitset_free (V1);
   bitset_free (P);
+  free (nterm_map);
+  nterm_map = NULL;
 }

src/reduce.h

@@ -32,6 +32,11 @@ bool reduce_nonterminal_useless_in_grammar (const sym_content *sym);
 
 void reduce_free (void);
 
+/** Map initial nterm numbers to the new ones.  Built by
+ * reduce_grammar.  Size nvars + nuseless_nonterminals.  */
+extern symbol_number *nterm_map;
+
 extern unsigned nuseless_nonterminals;
 extern unsigned nuseless_productions;
+
 #endif /* !REDUCE_H_ */